HELP! Hafler DH-200 misbehaving!
Colin B.
I'm an unhappy camper this week. Electronics seem to explode as soon as
I walk in the same room as them.
Anyways...
One channel on my Hafler DH-200 amp quit working yesterday. The first
thing I wanted to check was the fuse at the back, but in the process of
moving the amp around, I just about burnt my hand--the right heat sink
(the channel that died) was HOT!!!! Not just warm like usual, but REALLY
hot!
So I let it cool down for half an hour or so, and then checked the fuse.
It was fine.
Tonight I'm going to pop the lid and check things inside, i.e. both fuses
on the circuit board, wires shorting, etc.
Does anyone have any suggestions here?
As an aside, I've had no tweeter on one speaker for the past week or so,
but this occurred on the opposite channel. I can't imagine a different
load one one channel seriously frying the other one like this.
Thanks all,
Colin
Robert Morein
"Colin B." <cbi...@somewhereelse.nucleus.com> wrote in message
news:4068...@news.nucleus.com...
> Hey all;
>
> I'm an unhappy camper this week. Electronics seem to explode as soon as
> I walk in the same room as them.
>
> Anyways...
>
> One channel on my Hafler DH-200 amp quit working yesterday. The first
> thing I wanted to check was the fuse at the back, but in the process of
> moving the amp around, I just about burnt my hand--the right heat sink
> (the channel that died) was HOT!!!! Not just warm like usual, but REALLY
> hot!
>
> So I let it cool down for half an hour or so, and then checked the fuse.
> It was fine.
>
> Tonight I'm going to pop the lid and check things inside, i.e. both fuses
> on the circuit board, wires shorting, etc.
>
> Does anyone have any suggestions here?
>
outputs without damage.
Against that one must balance the reputation of the Hitachi MOSFETs as
having a higher failure rate due to internal defect, than bipolar output
transistors.
I would look at the MOSFETs, but you'll have to read up on how to test them.
Do NOT test them with a Lissajous box or other standard transistor curve
tracer.
The breakdown voltage of the gate oxide is about 15V, and standard
procedures would bust them.
These devices are far more sensitive to ESD than a typical MOS integrated
circuit, because they contain no input protection diodes.
Antistatic pad and wrist strap are mandatory.
Eric K. Weber
oscillating at VHF radio frequencies when the electrolytic dry up from old
age....
Also you could check the bias setting....
Rgds:
Eric
"Robert Morein" <now...@nowhere.com> wrote in message
news:bKGdnZlhBaH...@giganews.com...
Colin B.
> Does it still work but get very hot? ..... they have a nasty habit of
> oscillating at VHF radio frequencies when the electrolytic dry up from old
> age....
Nope. The channel that gets hot doesn't work at all. No output, nada.
> Also you could check the bias setting....
Just set it a few months back, to 250mA each side.
Colin
Colin B.
> "Colin B." <cbi...@somewhereelse.nucleus.com> wrote in message
> news:4068...@news.nucleus.com...
>> Hey all;
>>
>> I'm an unhappy camper this week. Electronics seem to explode as soon as
>> I walk in the same room as them.
>>
>> Anyways...
>>
>> One channel on my Hafler DH-200 amp quit working yesterday. The first
>> thing I wanted to check was the fuse at the back, but in the process of
>> moving the amp around, I just about burnt my hand--the right heat sink
>> (the channel that died) was HOT!!!! Not just warm like usual, but REALLY
>> hot!
>>
>> So I let it cool down for half an hour or so, and then checked the fuse.
>> It was fine.
>>
>> Tonight I'm going to pop the lid and check things inside, i.e. both fuses
>> on the circuit board, wires shorting, etc.
>>
>> Does anyone have any suggestions here?
>>
> That is a really tough amp. Rumored to withstand a direct short on the
> outputs without damage.
> Against that one must balance the reputation of the Hitachi MOSFETs as
> having a higher failure rate due to internal defect, than bipolar output
> transistors.
Thanks for the info Robert.
If I had a (or some) blown output transistors, could the heat sink get
hotter? My generic idea of failed parts is: dead part = open circuit =
no current flowing.
I didn't get to crack it open last night, but I'm hoping to find a small
bit of loose wire shorting the outputs, and no damage when its removed.
I might be living in a fantasy, though.
Colin
Robert Morein
They can fail either way, though it is true that open circuit is somewhat
more common, and that would not cause it to heat up.
You'll want to remove each supply rail fuse separately, and connect an
ammeter or a current measuring shunt. You want 200 ma quiescent bias current
on each.
Once you find which side is hot, you have a choice: start testing MOSFETs,
or look at the voltage on the MOSFET gates. If there's something out of line
there, you may be able to excuse the MOSFETs and look for a defect in the
driver circuitry.
Rich Andrews.
My guess is that a driver transistor failed and is driving the outputs
into DC coduction. Re-check your bias, I will be willing to bet it is way
off.
r
--
Nothing beats the bandwidth of a station wagon filled with DLT tapes.
Colin B.
> "Colin B." <cbi...@somewhereelse.nucleus.com> wrote in
> news:4068...@news.nucleus.com:
(One channel dead on Hafler power amp)
> My guess is that a driver transistor failed and is driving the outputs
> into DC coduction. Re-check your bias, I will be willing to bet it is way
> off.
Well, here's the latest:
The good channel has a bias of about 30mA when stone cold! (should be
250mA when fully warmed up--ten times increase seems odd). The bad
channel though, has a bias of about 900mA when cold. Yikes!!!
With no load attached, nothing warms up on the bad channel at all.
With speakers attached, I hear a 60Hz hum on the bad channel, and the
2SJ49s get ferociously hot in a matter of seconds.
As an aside, I've got a feeling I fried the unfused (10A max) ammeter
in my DVM. :-( Somedays I think I should just bury myself away from
anything electronic.
Anyways, do I have a hope for this thing, or is it a matter of replacing
the transistors ($80 a set!) vs. junking it?
And what would have caused it to die so suddenly on me?
thanks all,
Colin
El Meda
>Rich Andrews. <nw...@yihoo.com> wrote:
>> "Colin B." <cbi...@somewhereelse.nucleus.com> wrote in
>> news:4068...@news.nucleus.com:
>
>(One channel dead on Hafler power amp)
>
>> My guess is that a driver transistor failed and is driving the outputs
>> into DC coduction. Re-check your bias, I will be willing to bet it is way
>> off.
>
>Well, here's the latest:
>
>The good channel has a bias of about 30mA when stone cold! (should be
>250mA when fully warmed up--ten times increase seems odd). The bad
>channel though, has a bias of about 900mA when cold. Yikes!!!
>With no load attached, nothing warms up on the bad channel at all.
>With speakers attached, I hear a 60Hz hum on the bad channel, and the
>2SJ49s get ferociously hot in a matter of seconds.
Be careful: your amp has a heavy DC offset on the output of the bad
channel that could fry your speakers.
>Anyways, do I have a hope for this thing, or is it a matter of replacing
>the transistors ($80 a set!) vs. junking it?
Try disconnecting the driver transistors of the bad channel, and
measuring (with the DC Volts scale of your DMM) if there is DC on the
speaker terminals. If it's not, then probably the output transistors
are in good shape, and the trouble may be cheaper to repair.
---
Ing. Remberto Gomez-Meda <gom...@hotmail.com>
http://ingemeda.tripod.com/
INGE - Ingenieria Electronica.
Puerto Vallarta, Jalisco, Mexico.
Colin B.
> "Colin B." <cbi...@somewhereelse.nucleus.com> wrote:
>
>>Rich Andrews. <nw...@yihoo.com> wrote:
>>> "Colin B." <cbi...@somewhereelse.nucleus.com> wrote in
>>> news:4068...@news.nucleus.com:
>>
>>(One channel dead on Hafler power amp)
>>
>>> My guess is that a driver transistor failed and is driving the outputs
>>> into DC coduction. Re-check your bias, I will be willing to bet it is way
>>> off.
>>
>>Well, here's the latest:
>>
>>The good channel has a bias of about 30mA when stone cold! (should be
>>250mA when fully warmed up--ten times increase seems odd). The bad
>>channel though, has a bias of about 900mA when cold. Yikes!!!
>>With no load attached, nothing warms up on the bad channel at all.
>>With speakers attached, I hear a 60Hz hum on the bad channel, and the
>>2SJ49s get ferociously hot in a matter of seconds.
>
> Be careful: your amp has a heavy DC offset on the output of the bad
> channel that could fry your speakers.
Indeed it does! Luckily, the speakers are fine.
> Try disconnecting the driver transistors of the bad channel, and
> measuring (with the DC Volts scale of your DMM) if there is DC on the
> speaker terminals. If it's not, then probably the output transistors
> are in good shape, and the trouble may be cheaper to repair.
On the bad channel with no output transistors, I'm measuring -0.5VDC
output. That's pretty substantial. I measured the output with the
transistors in place, and it was sitting at -20VDC!
Next steps? It looks like the transistors might be OK.
Colin
El Meda
>> Try disconnecting the driver transistors of the bad channel, and
>> measuring (with the DC Volts scale of your DMM) if there is DC on the
>> speaker terminals. If it's not, then probably the output transistors
>> are in good shape, and the trouble may be cheaper to repair.
>
>On the bad channel with no output transistors, I'm measuring -0.5VDC
>output. That's pretty substantial. I measured the output with the
>transistors in place, and it was sitting at -20VDC!
>
>Next steps? It looks like the transistors might be OK.
>
>Colin
The driver transistors that I was talking about weren't the output
transistors, but the smaller transistors that are connected to them.
With the output transistors in place disconnect the smaller ones,
power on the amplifier, and check for DC on the speaker terminals. If
there is not DC, then the output transistors are probably OK.
Even If the output transistors are good, check the disconnected
drivers. Check in circuit (with the amplifier turned off) also every
diode and Zener on the bad channel, and every resistance of less than
1 Kohm. Check in circuit the even smaller transistors for short
circuits. Disconnect and test off-circuit everything suspicious.
Change the bad ones.
Colin B.
> The driver transistors that I was talking about weren't the output
> transistors, but the smaller transistors that are connected to them.
Argh! I can't even read. Where you said driver, I read output; and then
interpreted the rest of your post with that misinterpretation.
Anyways...
> With the output transistors in place disconnect the smaller ones,
> power on the amplifier, and check for DC on the speaker terminals. If
> there is not DC, then the output transistors are probably OK.
Will do this over the weekend. I'm visiting the parents, so my dad
gets to help out on this one. :-)
> Even If the output transistors are good, check the disconnected
> drivers. Check in circuit (with the amplifier turned off) also every
> diode and Zener on the bad channel, and every resistance of less than
> 1 Kohm. Check in circuit the even smaller transistors for short
> circuits. Disconnect and test off-circuit everything suspicious.
Thanks! I'll post with updates after the weekend.
Colin
Hafle...@ameritech.net.invalid
Q9 is a common weakness in this amp. It is part of the Vbe circuit that
controls the bias setting. Replace this transistor. Most any garden variety
NP2222 is OK, even an NPN from Radio Shack. Without a Variac your struggle
is probably not going to be successful, however.
Hafle...@ameritech.net.invalid
Hafle...@ameritech.net.invalid wrote:
Also, you have to understand that the most frequent problem with these DH-200
amps is DC offset, caused by imbalances amongst the 4 input differential amp
transistors. The NPN Q1/Q2 pair must be hFe matched within 10% or less. Ditto
for the PNP Q5/Q6 pair. The DH-220 included a small DC nulling circuit and
there is a mod for the DH-200 to add a simpler version of it. I have installed
a DC nulling circuit to two DH-200 amps and got DC offset down to less than 5
mV.
Driver transistors fail and output MOSFETs fail (rarely) but I would look
further upstream for your solution -- at least to begin with.
Dick
Robert Gault
><snip>
> Also, you have to understand that the most frequent problem with these DH-200
> amps is DC offset, caused by imbalances amongst the 4 input differential amp
> transistors. The NPN Q1/Q2 pair must be hFe matched within 10% or less. Ditto
> for the PNP Q5/Q6 pair. The DH-220 included a small DC nulling circuit and
> there is a mod for the DH-200 to add a simpler version of it. I have installed
> a DC nulling circuit to two DH-200 amps and got DC offset down to less than 5
> mV.
> Dick
>
Dick,
Could you send me a circuit diagram for this DC nulling mod for the
DH-200 by e-mail?
Hafle...@ameritech.net.invalid
Robert Gault wrote:
The DH-200 DC nulling circuit was described by Walt Jung in The Audio Amateur 1/83,
page 56. It is simple and mounts nicely in the circuit card by drilling 3 small holes
with a wire gauge drill (like for drilling holes in a PCB).
Just attach a 100Kohm pot between the two power supply rails and feed the output of
its wiper through a 2 megohm resistor to where R3 (or R3) touches the input trace. I
used a miniture 1/2 Watt multi-turn cermet variable resistor. I suppose a small film
cap across the legs of the pot would be a good idea and will do this soon and give it
an "ear" test. This little circuit solves a bunch of problems and I don't know why
more people don't use it. It is not a destructive mod, just a simple add-on.
The pot allows inserting through the 2 meg ohm resistor a very small "sample" of
either + or - voltage to the amp's input. I've seen more than one schematic for other
amps that routinely do the same thing.
Hope this helps. Spread the word!
Dick
Colin B.
> Hafle...@ameritech.net.invalid wrote:
>
>> Q9 is a common weakness in this amp. It is part of the Vbe circuit that
>> controls the bias setting. Replace this transistor. Most any garden variety
>> NP2222 is OK, even an NPN from Radio Shack. Without a Variac your struggle
>> is probably not going to be successful, however.
>
> Also, you have to understand that the most frequent problem with these DH-200
> amps is DC offset, caused by imbalances amongst the 4 input differential amp
> transistors. The NPN Q1/Q2 pair must be hFe matched within 10% or less. Ditto
> for the PNP Q5/Q6 pair. The DH-220 included a small DC nulling circuit and
> there is a mod for the DH-200 to add a simpler version of it. I have installed
> a DC nulling circuit to two DH-200 amps and got DC offset down to less than 5
> mV.
>
> Driver transistors fail and output MOSFETs fail (rarely) but I would look
> further upstream for your solution -- at least to begin with.
Well well well.
This weekend, we pulled and tested Q9. Sure enough, it was fried. One of
the leads from the power supply to the amp board itself was also broken
(cold solder joint, and melted insulation), so we fixed that.
The result? Success! Sweet, blissful music again! For about two hours,
after which it blew up again. This time it took the speaker output fuse
with it.
So it's not so much a question of Q9 failing, as much as something killing
Q9. Back to the schematics. Sigh.
Colin
Rich Andrews.
@news.nucleus.com:
Time for an oscilloscope I would say. I would check all of the resistors
around Q9 and any preceding drive stages.
Dick West
Colin,
I doubt anything Killed Q9. Read my previous comments (above) about the 4 input
transistors that make up the input differential amp. Did you monitor this channel for
DC offset? You must do this. DC offset will blow the speaker output fuse (to protect
the amp and speaker against the high DC offset). You must get your DC offset to be
less than 50 mV. Note that DC offset can change as the amp warms up.
AlsoI think you need to use a Variac and do some routine and tedious checking of
voltages, especially around these 4 input transistors. Check their voltages around
their CBE points. You need to correct the DC offset problem which is probably caused
by one of these 4 transistors not matching its mate. The oscillosope stuff can come
later to ferret out sneaky little problems with waveform analysis and distortion. It
is possible that a high frequency oscillation may be stressing the output fuses, but
I would first check those 4 transistors. Remember, they must be used in pairs matched
for hFe, else DC offset results. Unless you install the little circuit I described in
another previous message here.
My comments are based on lots of experience fixing these amps.................
Dick
John A. Weeks III
<Hafle...@ameritech.net.invalid> wrote:
> I would first check those 4 transistors. Remember, they must be used in pairs
> matched
> for hFe, else DC offset results.
Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.
-john-
--
====================================================================
John A. Weeks III 952-432-2708 jo...@johnweeks.com
Newave Communications http://www.johnweeks.com
====================================================================
Hafle...@ameritech.net.invalid
"John A. Weeks III" wrote:
> In article <40718999...@ameritech.net.invalid>, Dick West
> <Hafle...@ameritech.net.invalid> wrote:
>
> > I would first check those 4 transistors. Remember, they must be used in pairs
> > matched
> > for hFe, else DC offset results.
>
> Did the original poster ever remove the MOSFET's from the unit. If
> so, were they put back in exactly the same place? If not, it is
> possible that the matched sets were mismatched in the process.
> That would be enough to cause the problem that Dick is talking about.
>
> -john-
John, if I recall parts of this thread correctly, the OP said that he also found
one of the DC power supply wires to the circuit card was not attached. Obviously,
this would cause a huge set of problems such as he described.
Now, we do know that the two N-channel MOSFETs on a channel must be matched on Vge
within 10%, which is why each MOSFET is marked with the Hafler grading number.
Ditto for the P-channel MOSFETs. These devices are not the same 4 to which I
referred. I referred to the 4 input differential amp circuit on the circuit card
which must be matched, not the 4 MOSFETs attached to the heatsink.
Let's wait to hear more from the OP before making anymore guesses about what is
wrong. Besides, without the use of a Variac his efforts are probably doomed to
failure.
Dick
Colin B.
> Did the original poster ever remove the MOSFET's from the unit. If
> so, were they put back in exactly the same place? If not, it is
> possible that the matched sets were mismatched in the process.
> That would be enough to cause the problem that Dick is talking about.
Here I am, the original poster! :-)
I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.
And as for Dick's suggestions, let me recap and intersperse a few
questions.
1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms are:
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)
So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.
Then--MAYBE--I'll plug it in again.
Colin
Hafle...@ameritech.net.invalid
Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.
Here are some suggestions and opinions:
1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.
2. Dramatic changes in parameters that seem related to temperature could be
caused by a break (a lesion) in a trace, a solder joint, or even an internal
connection in an active device (transistor or diode) or a resistor. Things
expand and contract due to temperature and only a microscopic circuit break
is required to produce significant changes in output. But, without better
control in input AC voltage it is difficult to track down this type of
problem. You need a Variac. However a change in DC offset from 80 mV to 20V
is dramatic and makes me think you have an either/or thing going on, such as
a temperature related break in the circuit or a component.
Here is a repair strategy to consider:
1. Get a Variac or wire up 2-3 60 watt light bulbs in series with the input
AC. By switching a bypass to each bulb in turn you can slowly ramp up the
input AC surge and voltage -- giving you some control over events in the
circuit.
2. Use two meters, a Variac, a can of "cool-it" and a heat gun (hair dryer).
Monitor bias current and DC offset voltage with the two meters as you slowly
ramp up the incoming AC voltage. Put a 10 ohm 10 watt resistor across the
speaker outputs in the channel under test. This will help you more reliably
test for DC offset when you connect meter probes. And, make all voltage
measurements related to circuit ground -- the connection between the two
power supply filter caps.
Selectively heat and cool various transistors and monitor the effect of
these events by watching the changes in bias and, especially, the DC offset.
This will help you make some assumptions about which part of the circuit is
bad.
Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.
3. What color is your circuit card? Is it phenolic board or the gray-green
fiberglass? What is the serial number of your amp? It is on a 2" square
white sticker on the bottom or back side of your amp. There were 3 revisions
to the DH-200 and I can tell you which you have based on the serial number.
Earlier amps with phenolic cards often had microscopic dry solder joints
around the driver and pre-driver transistors (the 4 with the cooling fins).
This was because the solder bath temp and time this card could tolerate was
too low to produce good joints all of the time. Later versions of the amp
used a small plastic washer at the bottom of each of these 4 transistors and
the attachment of the traces on the rear of the card was improved (better
glue), which resulted in higher tolerances in solder bath time and
temperatures which resulted in improved solder connections without heat
damage to components.
As a minimum I would re-flow the solder joints around the BCE of these four
transistors, the ones with the small metal can tops.
4. And, of course, you should check for the obvious. Broken wires or bad
solder joints in other parts of the amp, especially power supply connections
and the feedback to the amp from the speaker protection fuse. However
because your problem is only with one channel I doubt anything is wrong with
the power supply unless it is a power supply connection to the card. You
could monitor the voltages at the + and - supply rails on the card as you
tap its components and its connections back to the PS. They should be ~65
+/- VDC.
5. You should lay out both heat sinks to enable access to both circuit
cards. Then compare voltages at suspicious parts by noting voltages on the
good circuit and the same spot on the bad card. But, of course, this
prolonged comparison of voltages works best when a variac is used so you can
keep the bad card at energized at a low voltage that precludes blowing up or
frying the rest of its channel.
Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.
Dick
Colin B.
> Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
> progress.
>
> Here are some suggestions and opinions:
>
> 1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
> modestly lower or higher hfe will result mostly in the inability of P1 to
> adjust the amount of bias current. You seem to be getting dramatic changes
> in voltages as a function of time (and probably temperature) too large to be
> related just to bias adjustment.
Agreed that Q9 isn't the offender--it does, however, seem to be a repeated
victim. Just to expand on my earlier comments...
When we pulled Q9, it was dead. It actually measured the same forwards
and backwards (i.e. BC and CB were identical in a diode tester, as were
BE/EB)
With the new Q9 in place and working, the DC offset at the speakers was
~80mV, and everything was fine. Then the channel died, the fuse blew,
and (after replacing the fuse) the DC offset was -20V. After this, letting
it cool or heat didn't have any effect. That is to say, the DC offset
seems to be caused by Q9 being blown open, not causing it.
So if it's a thermal problem, it's a thermal problem blowing up Q9. The
amp doesn't behave or misbehave based on temperature. (other than that
it took a few hours for the thing to blow up again--which could have been
either thermal, or pushing Q9 a bit harder than allowable, for those
hours, until it failed.)
In other words, there's no way of going back to a working amp, short
of replacing that transistor. The only way to actually FIX the amp, is
to find what's killing it.
The next time I see my parents, I'll grab my dad's variac and my old
'scope. I wish I hadn't been so optimistic on our repairs, and grabbed
the stuff when we were there last weekend.
> Also, use a plastic tool (toothbrush handle) to tap or gently strike various
> components and circuit traces to see if doing so changes the readings on the
> meters. This process may help you find any small circuit breaks that might
> exist. On one occasion I took the time to reflow all solder joints on the
> back of one circuit card that was causing a problem, and doing this fixed
> the circuit.
Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.
> Well, that's enough for now and I am sure enough to draw some flames from
> others. Tell me the serial number and I will make other suggestions.
I'll get that tonight.
By the way, thanks to you and all the others who have helped with this
amp. I really appreciate it!
Colin
Dave Platt
Colin B. <cbi...@somewhereelse.nucleus.com> wrote:
>Good point. I'm tempted to reflow the board as it is, just because.
>When we saw the one bad power lead on the board, we did touch up all
>of the connecting wires (none of them looked very good, actually) but
>not on the board itself. Speaking of which, it's a phenolic board on
>this amp. I don't have the serial number right now, but I'll get it
>when I go home. I'll look closely, but I don't think there are any
>washers between the transistors and the board, anywhere.
I had (have) a Van Alstine Omega 150 amp which developed some odd
symptoms - a sharp POP in one channel, anywhere from 15 seconds to a
couple of minutes after it was powered on. This occurred
intermittently for a couple of years, and then the channel suddenly
became extremely noisy (hiss).
I phoned AVA, and Frank Van Alstine came on the phone and explained
that the problem was of basically the same nature as has been
suggested for your Hafler amp - bad solder joints on some of the
connections between the driver transistors and the board. Repeated
thermal stress could eventually cause the joint to become intermittent
or open. It was apparently due to the soldering process used... AVA
changed the process once they learned of the problem with this batch
of amps.
He recommended that I reflow the individual solder joints and add a
bit of good-quality new solder to each. I did, and it fixed the
problem completely.
--
Dave Platt <dpl...@radagast.org> AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!
Colin B.
Oh well.
Here's the weekend's update on the amp.
First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.
Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.
So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!
I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.
Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.
Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.
So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?
Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).
So what next?
Thanks,
Colin
Mark D. Zacharias
a DC offset. This assumes a load was connected. You might start by adjusting
the bias with NO load. There could be a problem on the load side at this
point.
Mark Z.
--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.
"Colin B." <cbi...@somewhereelse.nucleus.com> wrote in message
news:4077...@news.nucleus.com...
Hafle...@ameritech.net.invalid
might see it, especially the part about a mod to fix DC offset in these amps.
Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.
Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.
At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.
Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:
A CIRCUIT MOD TO FIX DC OFFSET
-------------------------------------------------------
This DH-200 DC nulling circuit was described by
Walt Jung in The Audio Amateur 1/83, page 56.
It is simple and uses a small variable resistor that
mounts nicely on the circuit card by drilling 3 small holes
with a wire gauge drill (like for drilling holes in a PCB).
I drilled my 3 small holes in a vertical line beginning just
above where R8 is attached to the + PS rail.
The bottom leg of the pot was solder tacked to the + PS rail,
the other leg (top pin) is attached to the - PS rail by a short
length of insulated wire. The middle pin (wiper)
is attached to the junction of R4 and R5 via a
2 megohm 1/4 watt metal film resistor.
Just attach a 100Kohm pot between the two power
supply rails and feed the output of its wiper through
a 2 megohm resistor to where R4 and R5 touch the
input trace. I used a miniature 1/2 Watt multi-turn cermet
variable resistor. A small value film cap across the
legs of the pot would be a good idea to help cancel
any noise riding on the PS rails.
This little circuit solves a bunch of problems and
I don't know why more people don't use it. It is not a
destructive mod, just a simple add-on.
The pot allows inserting through the 2 meg ohm
resistor a very small "sample" of either + or - voltage
to the amp's input to null DC offset. I've seen more than
one schematic for other amps that routinely do the same thing.
Hope this helps. Spread the word!
Dick
==============================
Colin B.
> Pardon me for top posting but I want this material to be "up front" so others
> might see it, especially the part about a mod to fix DC offset in these amps.
No problem for me.
> Your soldering skills are excellent and it seems you are zeroing in on
> success. Did you state you still have a DC offset of 30 mV? Did you "load"
> that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
> the positive speaker terminal referenced to circuit ground? Don't be confused
> by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
> confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
> that part of the circuit.
First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.
As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).
> Your sr.# shows your Hafler kit was readied for shipment during the 13th week
> of 1981 and was a kit, not factory assembled.
Thanks much! Very cool to know.
Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!
Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.
> At this point I would suggest swapping the MOSFETs from one channel to
> another. It has been my experience that some of the older MOSFETs become
> "leaky" which could influence bias settings and cause extra heat. I've had
> some of the P-channel devices that no longer matched the original Hafler grade
> number and had gone down to a grade level of 1. Also, when checked for leakage
> they turned out to be too far out of spec to be usable. I have an original
> Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
> if the MOSFET swaps from one channel to the other make a difference in bias
> current and DC offset and heat sink temps.
I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.
> Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:
> A CIRCUIT MOD TO FIX DC OFFSET
This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!
So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.
Colin
Hafle...@ameritech.net.invalid
SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.
Dick
=========================
Colin B.
Hafle...@ameritech.net.invalid wrote:
> Colin,
>
> SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
> and the first 4 transistors are almost always the main culprits when too much DC
> offset is present. And, as you found, the solution is relatively simple for those
> who have good soldering skills.
Indeed, success at last! My soldering skills were a bit rusty, but have
come back with a vengeance. My next plan is to replace my crappy iron
with a nice new station (Weller WES51, or Hakko 936--any opinions folks?)
and then start putting all of my other projects back together.
As an aside, this amp may be up for sale before too long. Just a heads
up.
Thanks again, everyone who helped.
Colin
Hafle...@ameritech.net.invalid
Colin,
What about the temperatures of both the left and right heat sinks? Have they begun to
be similar? FWIW, I keep my Haflers of this vintage biased at around 325 mA. The
original factory spec was around 250 mA (conservative) per channel and many guys bias
theirs up to 450 mA per channel -- if the amp is kept properly ventilated. Supposedly a
Hitachi MOSFET reaches optimum operating area with 100 mA of bias so if you add up this
number for each of the 4 MOSFETs per channel and include around 50 mA to drive the
circuit card, you can understand how the 450 mA figure is derived.
However, the stock heat sinks are just not large enough for much of anything beyond 350
mA -- at least in my experience -- and I don't like the sound of a ventilating fan.
Dick the "HaflerLover"
Colin B.
Hafle...@ameritech.net.invalid wrote:
> What about the temperatures of both the left and right heat sinks? Have
> they begun to be similar? FWIW, I keep my Haflers of this vintage biased
> at around 325 mA. The original factory spec was around 250 mA
> (conservative) per channel and many guys bias theirs up to 450 mA per
> channel -- if the amp is kept properly ventilated. Supposedly a Hitachi
> MOSFET reaches optimum operating area with 100 mA of bias so if you add up
> this number for each of the 4 MOSFETs per channel and include around
> 50 mA to drive the circuit card, you can understand how the 450 mA figure
> is derived.
I was running mine at 240mA before the repairs, and then set them to
220mA since repairing everything. The heatsinks are close enough that
I can't be sure by touch if they're different or not--the right one
(the repaired channel) might be SLIGHTLY hotter, but it's hard to say.
I think I'll bump the bias up to 300mA and see how hot it is for our
house. If nothing else, the increased bias should get the thing up
to operating temperature faster, when first turned on 'cold.'
Robert Gault
><snip> FWIW, I keep my Haflers of this vintage biased at around 325 mA. The
> original factory spec was around 250 mA (conservative) per channel and many guys bias
> theirs up to 450 mA per channel -- if the amp is kept properly ventilated. Supposedly a
> Hitachi MOSFET reaches optimum operating area with 100 mA of bias so if you add up this
> number for each of the 4 MOSFETs per channel and include around 50 mA to drive the
> circuit card, you can understand how the 450 mA figure is derived.
>
> However, the stock heat sinks are just not large enough for much of anything beyond 350
> mA -- at least in my experience -- and I don't like the sound of a ventilating fan.
>
> Dick the "HaflerLover"
>
In effect you are saying that the Hafler design is flawed as 450ma is
grossly different from 200-250ma. You may be correct but what objective
testing have you done to verify that any bias different from the Hafler
spec is better?
Have you performed THD or IM distortion tests at 450ma vs 200-250ma or
any other type of test other than listening? How, if at all, does the
change in bias affect the power rating or other specs as listed in the
Hafler manual?
Hafle...@ameritech.net.invalid
Robert G,
No, I am not saying the design is flawed. The heat sinks could be a little larger, but this
amp was designed to meet several price points, it was never portrayed as a cost-no-object
state-of-the art amplifier. Biasing of Hitachi MOSFETs and the resultant sound seems to exist
on a contiuum and the observed sound difference between a 300mA and 400mA bias per channel is
subtle, at best. Time and again "experts" comment about the subtle improvement in sound as
these devices are biased at a higher current level. Noever, in any of the many articles I
have read over the last 20 years were deleterious effects mentioned, such as the THD or IM
measures you mention.
Personally, I have not run the tests you describe and I really wish you or someone competent
would do so. In the meantime I will ask you if you are familiar with the Audio Amateur
article (4/1981) by Walt Jung and Dick Marsh on the mods for this amp? These authors, very
competent and well known EEs and circuit designers, state the amp should be biased at 400-425
mA per channel. Contact John Hillig at Musical Concepts and he will state that these devices
sound best when biased up to 100 mA each or 400 mA per channel. Mr. Hillig is a competent
circuit designer.
I realize that these "golden ear" type comments are anathema on this technically oriented
newsgroup, but science and art have to coexist, don't you agree? If you can bring to us any
of the type of objective "proof" you mention I believe we all would be grateful.
Dick
Colin B.
> In effect you are saying that the Hafler design is flawed as 450ma is
> grossly different from 200-250ma. You may be correct but what objective
> testing have you done to verify that any bias different from the Hafler
> spec is better?
>
> Have you performed THD or IM distortion tests at 450ma vs 200-250ma or
> any other type of test other than listening? How, if at all, does the
> change in bias affect the power rating or other specs as listed in the
> Hafler manual?
Don't know about Dick, but it seems to me that the bias would help
eliminate crossing distortion, wouldn't it? If so, it should be pretty
easy to measure on a 'scope.
As for measuring THD, I've got a lovely old distortion analyser in the
basement that tops (bottoms?) out at 0.1% THD. Anyone want to buy it?
:-)
Colin
Michael
point is apprx 100mA per device. That's the main reason the 100mA number is
so "popular". It's actually a different for the N and P channel devices but
100mA per device is close enough for rock and roll, as they say :-)
Currently, I am using somewhat similar lateral MOSFETs manufactured by
Toshiba in my amp (circuit is kinda/sorta similar to a later Hafler
product), biased at over 300mA per device with 4 devices in parallel per
rail per amplifier. I've got bigger heat sinks than the Hafler DH200,
though. And I don't use +/-65V rails. I'm driving 4 ohms loads - Kef 104/2
speakers.
Very nice indeed :-)
Michael
Hafle...@ameritech.net.invalid wrote in
news:407EA7B1...@ameritech.net.invalid:
Colin B.
> The datasheets for the Hitachi Power MOSFETs indicate the zero tempco bias
> point is apprx 100mA per device. That's the main reason the 100mA number is
> so "popular". It's actually a different for the N and P channel devices but
> 100mA per device is close enough for rock and roll, as they say :-)
Well, since this amp is open on my bench and staying away from my stereo
for quite a while, I plan on playing with it quite a bit. I'm going to
hook up a signal generator and a 'scope, and see how everything gets
created.
Should be fun, at least.
> Currently, I am using somewhat similar lateral MOSFETs manufactured by
> Toshiba in my amp (circuit is kinda/sorta similar to a later Hafler
> product), biased at over 300mA per device with 4 devices in parallel per
> rail per amplifier. I've got bigger heat sinks than the Hafler DH200,
> though. And I don't use +/-65V rails. I'm driving 4 ohms loads - Kef 104/2
> speakers.
So 16 MOSFETs pulling 300mA each at idle? I'm glad I don't have your
power bill! :-)
Colin