Nick Desmith smps board question

[RHC]

Hi all

I have been playing with an smps a friend built which is supposed to be identical to Nick Desmith's board 

http://desmith.net/NMdS/Electronics/NixiePSU.html

with the exception that its a through hole board rather than smd.  Im an audio guy btw not a nixie builder.  With a 12ax7a at 300v/1ma it works well. Noise free, stable, sounds great.  But when I tried it with a circuit that drew 20ma of current, both the voltage and current output collapsed to 78v/4ma.  I tried changing the resistor that sets the output voltage to 200v and got similar results.  Im using a 9v 3amp wall wart supply.  I know its impossible to really diagnose anything without a circuit which is the next thing Im going to come up with, but I thought it couldn't hurt to post this and see if there is anything obvious those of you much more familiar with these supplies than me, might suggest as possibly the problem or what to look for. 

thanks!

Bob

[Arne Rossius]

Hi,

RHC wrote:
> http://desmith.net/NMdS/Electronics/NixiePSU.html
> with the exception that its a through hole board rather than smd.

Since your friend apparently didn't use the exact same PCB layout, I
think some photos of your board might be helpful as well to diagnose the
problem. With the MAX1771, it is very important that the ground
connection between the IC and the current sense resistor is very short,
and that little current flows through it. It's a good idea to feed the
input ground to the sense resistor first, and then add another trace
directly from the resistor to the IC. The same goes for the CS pin
connection, it should be connected as close to the resistor as possible.
Are you sure that your circuit actually has a 50 Milliohm resistor? I
haven't often seen values this small as through-hole components.

> Im using a 9v 3amp wall wart supply.

Have you tried using a higher input voltage, e.g. 12V or 15V?


Best Regards,
Arne

[John Rehwinkel]

> I have been playing with an smps a friend built which is supposed to be identical to Nick Desmith's board
>
> http://desmith.net/NMdS/Electronics/NixiePSU.html
>
> with the exception that its a through hole board rather than smd. Im an audio guy btw not a nixie builder. With a 12ax7a at 300v/1ma it works well. Noise free, stable, sounds great. But when I tried it with a circuit that drew 20ma of current, both the voltage and current output collapsed to 78v/4ma. I tried changing the resistor that sets the output voltage to 200v and got similar results. Im using a 9v 3amp wall wart supply. I know its impossible to really diagnose anything without a circuit which is the next thing Im going to come up with, but I thought it couldn't hurt to post this and see if there is anything obvious those of you much more familiar with these supplies than me, might suggest as possibly the problem or what to look for.

Does the switching FET get hot? If so, you probably need a lower resistance one. What's the DC current rating on the inductor?

The efficiency and effectiveness of these single-inductor boost supplies starts to really fall off at large voltage ratios (like you'd have with 300V out). You might consider going to 24V or more on the input side (with appropriate modifications to correctly power the control chip), using a transformer, or adding a voltage doubler.

- John

[RHC]

I think I may have figured out the problem based on John's comment. This is from Nick Desmith's web page. 

• For 50mA+ output, L1 should be DC rated at about 2A. The chosen inductor, an EPCOS B82479, fits this bill and works very well in this circuit. If you use a shielded inductor, you will get less RFI, but the efficiency will also drop by about 3%. Small surface mount, high current, shielded inductors are not common! You could try a Sumida CDRH127-101 which works ok even though it starts to saturate at about 1.7A.
• For lower output currents up to 25mA, you can use a 1A inductor and change R_sense to 0R100 Ω. A good inductor here would be a Sumida CDRH125-101. Select an inductor with a low DC resistance as straight resistive losses will effect efficiency and lead to heating.

Sure enough my board is using the 1A Sumida.

Also I notice my friend used a different vishay FET, I assume to get the 300v (I only need 200)  Nick specified this one. 

http://www.vishay.com/docs/91039/91039.pdf

My friend is using this one. 

http://www.vishay.com/docs/91043/sihf720.pdf

I notice that the RDS on value is higher on the 720.  1.8ohms instead of .28 ohms.  Could this be contributing to the problem?

Last question.  From reading Nicks webpage: Switcher rate was approximately 62kHz

This doesn't seem to be a problem but I was wondering since this is an audio application if there would be a lot involved in moving this up to 100 -120K or even more and still get 200v/50ma out. 

I really appreciate all your help. 

Bob

[John Rehwinkel]

I notice that the RDS on value is higher on the 720.  1.8ohms instead of .28 ohms.  Could this be contributing to the problem?

That's about seven times as much, so with a higher current inductor, it'll start to impact your efficiency.

Last question.  From reading Nicks webpage: Switcher rate was approximately 62kHz

This doesn't seem to be a problem but I was wondering since this is an audio application if there would be a lot involved in moving this up to 100 -120K or even more and still get 200v/50ma out.

The main problem with higher frequency switching is the FET turn on/off time.  The higher the frequency, the shorter the period, and therefore the larger proportion of the time you'll be spending

turning the FET on and off.  During the turn on/off period, the FET is in its linear region and is dissipating more power.  Ideally, the FET is all the way off (dissipating no power) or all the way on

(dissipating only the power from its RDS on resistance times the current flowing through it).

The limiting factor in how fast you can turn the FET on and off is its gate capacitance divided by the available gate drive current.  To speed up the switching time, you can either choose a FET

with a small gate capacitance, or increase the gate drive current.  To increase the gate drive current, you can either build a transistor booster or use a dedicated FET gate driver chip such as

a MAX627.

- John

[RHC]

John 

thanks for the follow up, you have really been helpful.  Im sure what specific specs gate capacitance and gate drive current are.  So for example on this spec sheet 

http://www.vishay.com/docs/91039/91039.pdf

is gate capacitance = input capacitance 1300pf

and im not sure what gate drive current = ?

Bob

[gregebert]

For a MOSFET, there is no static drive-current except for a few nanoamps of leakage, which can usually be ignored.

The gate-drive-current is whatever is needed to slew the gate-source capacitance (aka input capacitance).

Using i=Cdv/dt, you can calculate how much drive-current your gate-driver needs to provide. From the datasheet, you need at least 4.5V for the gate-voltage.

If you want a turn-on time around 250nsec, you would need about 25mA. Remember this is not continuous current; it's just there when turning the MOSFET on or off.

Take note of the gate-charge graph (Fig 6 on the datasheet) if you are using more than 5V for the gate; it's a non-linear capacitance which is typical for MOS devices.

If you really want to do an accurate assessment, a SPICE simulation is best though that might be overkill in this application.

You can get-by with a smaller gate-driver, but that will increase your switching loss; it's a tradeoff.

[RHC]

Since I put up this thread I have since upgraded the coil, fet and diode but Im still having the problem.  With 9v/3a power supply the output drops from 300 volts no load to 70 volts at around 45ma. 

I was wondering if anyone might comment of this:

https://docs.google.com/spreadsheets/d/1KDNKAQ2sZJch5CYoSZMCEc1vGZVVrP8h2sKTUL66Opw/edit?usp=sharing

If they see anything that might be causing the the problem. 

Thanks!

Bob

[gregebert]

First thing to do is get a scope and start probing around. If you're measuring with a meter, it wont show the peak voltage.

I recently debugged a nasty switching-regulator problem in my wristwatch, only to discover the bench supply plus the inductance of the testleads I was using had tons of noise, so check all around the switching regulator as well.

Also be aware the feedback path to the regulator has a low-pass filter with a cutoff frequency around 160Hz (assuming I did my math correctly for Fc=1/2pi*R2*C2. You can ignore R1 because R1>>R2). That will definitely limit the response-time of the regulator. As you increase the load current, the converter will need to respond faster to replenish the output voltage (hint: I=Cdv/dt ). Your measurements clearly show the output voltage dropping at higher load current, so that's a definite clue.

Also, if you can run simulations that will help a lot. LT Spice (Linear Technology) is a great tool; I dont know if they have a competing part for the Max1771 (Maxim ?), or if Maxim has a similar tool.

[Nick]

On Wednesday, 15 April 2015 23:18:17 UTC+1, RHC wrote:

I was wondering if anyone might comment of this:

https://docs.google.com/spreadsheets/d/1KDNKAQ2sZJch5CYoSZMCEc1vGZVVrP8h2sKTUL66Opw/edit?usp=sharing

The MAX1771 is really sensitive to layout as well as component choice, as I mention on the web page. The schematic in the above link is certainly not my original schematic (see attached).

You may well have instability on the feedback line and HF oscillation with the board layout as is - do you have a decent 'scope?

Nick

[taylorjpt]

You are probably saturating the inductor since you don't have a current sens resistor in the source connection of the output transistor.

jt

[gregebert]

Another tip: Probing high-frequency signals, especially when there are nearby sources of electromagnetic energy, is tricky.

There are several good articles online, depending upon the type of probe you use. If you're lucky (I'm not...) you'll have a differential probe. Otherwise, a 10X probe is a must.

BEWARE of ground leads on probes! They will pickup noise. The best solution I found is to remove the ground clip and the plastic probe shield, which should expose the metal ground connection on the probe. Tightly wrap bare-wire around the ground shroud, then solder to the nearest circuit-ground. I was able to reduce the ground connection length from 6 inches to less than 1/4 inch, and it made a big difference. For the signal connection, I used bare wire-wrap wire, wrapped a few turns on the probe-tip, then soldered to the component being probed. The length was around 1/2 inch.

Yes, it's a pain, but there's nothing worse than spending (wasting!) weeks chasing-down a suspected noise problem only to find out you created it with improper measurement techniques.

Also be careful of non-obvious ground loops. The signal ground of your scope is likely to be connected to the AC power-line ground. If your circuit ground is connected to the AC line ground (either directly or through your power supply), you will likely have aground loop, and that will add noise.

[taylorjpt]

Here's a PDF from 2004 on probe connections:  http://www.tayloredge.com/reference/Electronics/TestBench/ScopeGrounding.pdf

Especially when testing power supplies the ground wire acts as a single turn magnetic pickup.

[RHC]

Thanks a ton for comments!  The board as is, is stable and cool and apparently noise free, and has run 1 12AX7 at 300 volts with a 9v supply in the field for long periods of time in a audio application. Of course 1 12AX7 draws very little current.  I dont know why RSense was left off but thats the next thing im going to try.  And I will find someone with a bench capable of doing a proper high frequency oscillation test. 

In the meantime I going to try a couple of Taylor edge boards for this new project, assuming the switching freq. is high enough for an audio application. 

Thanks again, if you folks think of anything else to check please post! :) 

Bob