MPSA42 or MJE340?

[Paul Andrews]

Go...

[jf...@my-deja.com]

I would say whichever is cheaper, keeping in mind that nothing is cheaper than free from the junque box.  I used the floor sweepings  (advertised as npn 300V and 10W) from Poly Paks, and none of these have failed yet.  I did have one genuine Motorola MPSA92 fail after 40 years of service.  Just make sure you are driving your transistors into saturation.  Obviously TO92 is smaller than TO220, and this might be a factor in your application.

[Mac Doktor]

On Apr 22, 2020, at 9:39 AM, 'jf...@my-deja.com' via neonixie-l <neoni...@googlegroups.com> wrote:

I would say whichever is cheaper, keeping in mind that nothing is cheaper than free from the junque box.  I used the floor sweepings  (advertised as npn 300V and 10W) from Poly Paks, and none of these have failed yet.  I did have one genuine Motorola MPSA92 fail after 40 years of service.  Just make sure you are driving your transistors into saturation.  Obviously TO92 is smaller than TO220, and this might be a factor in your application.

The perfect example of this is driving an IN-13 vs an IN-9. The former will work with an MPSA42 but the latter requires an MJE340 because of the higher current. It's well known that the bargraph tubes work more reliably if they're strobed; all of the circuit examples in Russian data sheets for the IN-9 and IN-13 use a half-wave rectified 220V mains power source. And now I finally understand a circuit modification I saw in a forum thread that has puzzled me for some time. The OP couldn't get an IN-9 (using an MJE340 of course) to work reliably using a generic barpgraph driver circuit. One of the replies added a 555 running at 50Hz driving a MPSA42 in series with the MJE340. What? How could you use the smaller trans... Now I get it. The MPSA42 was being used as a switch. DUHH.

Thank you for that insight. Cue up the next great mystery of life...

Terry Bowman, KA4HJH
"The Mac Doctor"

https://www.astarcloseup.com/

"Every kid starts out as a natural-born scientist, and then we beat it out of them. A few trickle through the system with their wonder and enthusiasm for science intact."—Carl Sagan, Psychology Today, 1996

[Paul Andrews]

Both transistors are rated for 500mA, so I don’t see why you need a MJE340 for an IN-9. This is particularly relevant for me as I want to switch around 50mA using 3.3V. Not sure what the voltage is right now, given that this is the cathode.

On Apr 22, 2020, at 5:25 PM, Mac Doktor <themac...@gmail.com> wrote:

--
You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/_TX-Z5A27JQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/B30CB5AB-765F-4F1C-AC1C-8F7D176739EE%40gmail.com.

[jf...@my-deja.com]

What do you have against the MJE340?  Is it too large?  Is it too much more money (onesies retail in Silicon Valley is 34 cents versus 75 cents)?  If I had to go out and buy new parts in hobby quantities, I would get the MJE340 and not invest any more brain power, but if it was going to be used in quantities of thousands I would choose more deliberately.  If I already had some MPS-A42's on hand, I would wire it in to the circuit and test to see how hot it gets, and switch to the MJE340 if it got noticeably warm. Oor you could do the Spice modeling and formal thermal analysis.  Actually, it would not surprise me if you opened them up and found that they used the same chip and differed only in the massiveness of the packages.

[Mac Doktor]

On Apr 22, 2020, at 9:03 PM, 'jf...@my-deja.com' via neonixie-l <neoni...@googlegroups.com> wrote:

What do you have against the MJE340?

Me? Nothing. All of the IN-9 projects that I've seen online use the MJE340. 

I'm a duffer and I'm here to learn. Paul wanted everyone to pile on so have at it! I'd be glad to hear more.

Terry Bowman, KA4HJH
"The Mac Doctor”

"Never install version point-zero of anything"

[Paul Andrews]



Or me? Well, the MJE340 is in a larger package, but other than that, nothing. What Terry said, though, is that the IN-9 requires a MJE340, not a MPSA42 because you need to sink more current. As far as I can tell, 10mA is well within the current handling capabilities of both transistors, so it is interesting that all the designs for the IN-9 use the MJE340, whereas all the designs for regular nixies use the MPSA42.

I happen to have an IN-9 spectrum analyzer sitting on a breadboard that uses the MJE340. I swapped them out for MPSA42 and the results definitely aren’t as good. To stick the glow to the bottom I periodically pull the base to ground - so turning the tubes off. With the MJE340 this works fine. With the MPSA42, not so much. So either they aren’t turning off fully, or they aren’t responding fast enough. Temperature-wise, they were barely above ambient. YMMV.

--

You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/_TX-Z5A27JQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/33F621FB-E6D6-42FB-AC61-146B81CAC0A7%40gmail.com.

 

[gregebert]

There are thermal considerations; the MJE340 is designed to be used on a heat sink and is rated for 20W, vs 1.5W for the MPSA42.

Also, there are some differences in the safe operating area curves; the MJE340 can tolerate a bit more current at any given voltage.

MJE device also has a lower Vce (sat) for a given collector current.

That robustness doesn't come for free; the MJE device has 1000x higher leakage current.

So, as with any design, you have to look at all of the datasheet parameters carefully.

[jf...@my-deja.com]

So this is a matter of academic curiosity and not an effort to make your device work.  I think this is a neat thing to do.  Based on your observations, this suggests some experiments.

 

In a previous post, I speculated that both part numbers might use the same chip.  I now think this is less likely.  Since this would be a destructive test, this should be done after all the other experiments.  You can expose the silicon chip by using wire cutters to scrape away the epoxy.  The chips will usually be bonded to the collector lead.

 

If they do use the same chip, the difference could be due to a higher temperature in the smaller package.  I would try using some canned Freon to chill the MPS-A42 to see if this changed its performance.

 

Another less likely possibility is that this is the natural variation in the chip processing.  You test this by comparing several parts with different date codes.

I am sure that creative minds can conjure up other hypotheses.

[gregebert]

I'm absolutely certain the MJE die is physically larger. For one thing, it must conduct more thermal energy to the case.

Another tip-off is the difference in DC current-gain, and that in-turn accounts for part of the 1000x difference in leakage current.

I believe leakage current is directly proportional to the area of the base (my device physics textbook is at work, and I'm at home recalling a class I took in 1983...); the actual leakage-current density is an exponential equation based on all sorts of properties of the silicon.

[Dekatron42]

Central Semiconductor Corp list them as having the same die size, check their documents named "Device datasheet" here https://my.centralsemi.com/product/partpage2.php?part=CP310-MPSA42 and here https://my.centralsemi.com/product/partpage2.php?part=CP310-MJE340, now I don't know if other manufacturers have the same data on theirs, but at least one manufacturer seems to use the same die in both types (maybe they will confirm if you ask them?).

/Martin

[jf...@my-deja.com]

On Thursday, April 23, 2020 at 3:10:36 PM UTC-7, gregebert wrote:

> I'm absolutely certain the MJE die is physically larger. For one thing, it must conduct more thermal energy to the case.

I am inclined to agree, but if it were me, I would be willing to invest $1.08 and a half hour in the experiment.

> Another tip-off is the difference in DC current-gain, and that in-turn accounts for part of the 1000x difference in leakage

> current.  I believe leakage current is directly proportional to the area of the base (my device physics textbook is at work,

> and I'm at home recalling a class I took in 1983...); the actual leakage-current density is an exponential equation based

> on all sorts of properties of the silicon.

All other things being equal, I would expect leakage to be roughly proportional to the chip are, which would not account for the 1000X.  Also, some specs in the data sheet are very loose, like guaranteed maximum leakage.  If the base and emitter are both diffused, I would expect these variations to swamp out the area.  How do they get the HV capability?  If it is an epi base layer, there would be variations due to the quality of the epitaxy that could affect the leakage.  That is a long way of saying that you might be right, but I don't know.

[Dekatron42]

The data differs though.... I just read that, missed it at the first look.....

/Martin

[gregebert]

Much of the additional leakage current is caused by the higher DC gain of the MJE device; the datasheet says gain peaks around 240.

Basically, the reverse-leakage current from the collector-base junction gets amplified by the DC gain (beta), to produce the overall leakage (Icbo)

[jf...@my-deja.com]

Is there a 1000X difference in the gain?  Across all lots and all operating conditions, I would have expected less than 10X.

[gregebert]

It's the leakage (Icbo) that is 1000x greater in the MJE device, not the current gain. There are many factors affecting leakage and current gain:

• Some are environmental (temperature, reverse-voltage, operating current)
• Some are device construction (physical dimensions of base width & area; emitter area), processing (diffusion, epitaxial growth)
• Some are chemical (doping levels for N & P)
• Some are device physics(leakage current density, density of states, etc. Lots of math here..)

The leakage is also directly proportional to the DC current gain.

The nice thing about nixie tubes is they use relatively low current, so most devices used for drivers wont get pushed into unsafe areas of operation.

[jf...@my-deja.com]

Sure, but based on the similarities in the Central Semi data sheets, the same die appears to be used interchangeably.  So the factors I can imagine is that they select and direct the leakier parts to the TO-220 package, or that the TO-220 package makes a difference (like allowing a higher junction temperature before the chip will self-destruct, or an intrinsically leakier package), or that the MJE is allowed to be exposed to more adverse circuit conditions (and they would have the same leakage in the same circuit).  Otherwise, I lack the imagination to conjure up other plausible differences as long as they use the same chip.  Or maybe other manufacturers do not use the same chip for both part numbers, and Central decided that using their MPS-A42 chip would not violate the Icbo promises for the MJE-340 (from Motorola since Central is a second source).  If I had parts from several companies, I would be even more motivated to open them up and compare.

[gregebert]

When I was in high school, I got to use a Tek 575 curve tracer. Perfect instrument to nail-down any differences between these devices.

Every now and then, I take a look on Ebay and ask myself if I should buy one. Yikes...I just took a peek and there are several available. Fortunately I just bought 2 more tubes from Dalibor so I've zapped my budget for now.

Anyways, if someone has a curve tracer it would be really interesting to compare these.

[jf...@my-deja.com]

You might have to settle for one of the more modern transistorized curve tracers.  In the last century when 500-series scopes were going for $20-$50, I remember talking to someone at one of the ham swaps.  He said that they were being bought up and cannibalized by the audiophools because they were full of 12AX7s and other coveted tunes.

[Instrument Resources of America]

I have a couple of 575 that I might sell, either as is, or reconditioned and calibrated (not inexpensive), with an operators manual. I also have the 575 mod 122C which sports a 400V collector sweep power supply. The audio people are crazy over them because they can be converted for TUBE curve tracing. If interested contact me OFF of this page.   Ira.

On 4/24/2020 8:41 PM, 'jf...@my-deja.com' via neonixie-l wrote:

You might have to settle for one of the more modern transistorized curve tracers.  In the last century when 500-series scopes were going for $20-$50, I remember talking to someone at one of the ham swaps.  He said that they were being bought up and cannibalized by the audiophools because they were full of 12AX7s and other coveted tunes.

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/c1dc59dc-cd04-4a21-818f-d331f6f7dc8c%40googlegroups.com.

[gregebert]

As much as I would love to have my own curve-tracer, I have to be practical. The transistors I select have SPICE models, so I simulate everything in detail.

Nixie tubes, especially the rare or expensive ones like b7071 and R|Z568 , I generate I-V curves by hand for each segment/cathode. A curve-tracer would help if it could export data to Excel. If I see failure or degradation later on, I retake the I-V data and compare it to the original measurements.

[jf...@my-deja.com]

It should be easy to convert the 575 back into a tube curve tracer since it was a modified 570 with the filament transformer removed and the sockets changed.  If you can find a meterless Hickok tube tester, you have the sockets, switches and filament transformer you would need to convert it back for a single tube.  It can obviously be done by brute force, but I have not figured out a clever way to set it up to do A-B comparisons so you can find matched pairs.  Some other things you might have to add are power supplies for the screen and suppressor and the other elements you might find in a pentagrid converter (or you could derive these from the voltages applied to the other elements).

[Mark Moulding]

I had a 575 - two, actually; it's quite a chunk of gear, weighing in at 70 pounds, with 39 vacuum tubes plus the CRT.  I picked them up at an auction, fiddled with them a bit, and decided I had other better uses for the space.

But a transistor curve tracer would still be a handy thing, and on eBay I stumbled across the Heathkit IT-1121, which has most of the functionality of the Tektronix unit without the oscilloscope (and a *lot* smaller).  To use it, you connect it to your own oscilloscope, which really needs to be an analog 'scope in X-Y mode; digital 'scopes won't work very well.  It does the multiple simultaneous curves and everything, just like the Tektronix.  It takes a little time to calibrate the oscilloscope each time, so I have an old HP unit that I just leave set up for the curve tracer.

There's currently one on eBay listed for parts only, which would probably actually be OK once the knobs were tightened back up. The circuitry inside is really quite simple, and there's not much that can go wrong with it.  Heathkit also had a model IT-3121 that I think is basically the same unit.  I also see listed a Leader LTC-905 which looks to be a very similar type of instrument.

If you want to trace vacuum tube curves, head over to https://www.dos4ever.com/uTracer3/uTracer3_pag2.html  The very nice gentleman there makes an entirely solid-state vacuum tube curve tracer as a kit, controlled by a pre-programmed PIC.  Rather than doing all the steps and curves simultaneously, it steps through the process, so it just needs a little switching supply rather than a big complex system.  It's very flexible, and the associated PC software (required, but free to download) really makes it easy.  No affiliation, just a satisfied customer.

~~
Mark Moulding

[Dekatron42]

You can also turn the uTracer into a transistor curve tracer: https://www.dos4ever.com/uTracerNotebook/Notebook.html#tortester1 , I haven't done it but thought about it a few times.

There's also the new uTracer V6, not a kit yet but it has a weblog: https://www.dos4ever.com/uTracerlog6/tubetester6.html

/Martin

[Paul Andrews]

Or how about BS107P instead? Same job, no base resistors.

BTW I was trying to use a SN75468N to control the IN-28 panel in the video below. It failed miserably. Segments that were supposed to be off were being pulled down to about 20V. Seven BS107P are working nicely, and can still be controlled using 3.3V.

https://youtu.be/qXi3FdB2HFw

[gregebert]

I suspect that leakage current is causing the Darlington driver (2-stage high current-gain amplifier) to turn on. With a base resistor (see datasheet), any leakage current into the base can still get amplified, because some of it will leak into the Darlington-pair and get amplified. A scope will give you a good idea what the leakage looks like; could be a series of "glitches" when the bulb turns on for a short time.

With no base resistance, all leakage-current into the base gets shunted away so it never gets amplified.

How high is your VCC supply for the neon bulbs ? It's certainly well-above the 100V spec for the driver, and despite the voltage-drop across the tube, there are occasional higher-energy electrons that can go thru the tube with a lower-than-normal voltage drop and they will greatly increase the leakage current.

[Paul Andrews]

VCC is 240V. Anode obviously sees less, plus I tried clamping the Darlington outputs to, say, 70V with limited success. The voltage divider I used for this needed to have low value resistors to have any noticeable effect, so quite a lot of current through the divider, except no measurable current into the COM pin.

All in all, thoroughly confusing, but  I’m happy with the BS107P.

I’ve run out of bench space for my old Tektronix scope, I need to get a more up to date one.

On Apr 29, 2020, at 12:12 AM, gregebert <greg...@hotmail.com> wrote:



I suspect that leakage current is causing the Darlington driver (2-stage high current-gain amplifier) to turn on. With a base resistor (see datasheet), any leakage current into the base can still get amplified, because some of it will leak into the Darlington-pair and get amplified. A scope will give you a good idea what the leakage looks like; could be a series of "glitches" when the bulb turns on for a short time.

With no base resistance, all leakage-current into the base gets shunted away so it never gets amplified.

How high is your VCC supply for the neon bulbs ? It's certainly well-above the 100V spec for the driver, and despite the voltage-drop across the tube, there are occasional higher-energy electrons that can go thru the tube with a lower-than-normal voltage drop and they will greatly increase the leakage current.

--
You received this message because you are subscribed to a topic in the Google Groups "neonixie-l" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neonixie-l/_TX-Z5A27JQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/e6f2268a-78d8-4a3d-af13-eef837bbdb5e%40googlegroups.com.

[jf...@my-deja.com]

As a matter of personal taste. I like to have a generous margin on BVceo, like always larger than Vcc.  To me, anything else is penny wise and pound foolish, especially in hobby quantities.  I always use transistors rated at 300V even if my Vcc is less than 200V.

Gregenert is probably right about amplified leakage, but another possibility is that you have avalanche breakdown of the CB junction in the driver transistor.  Given the nonlinear IV of neon, this might mean that your lamps are off when you first apply power, but it will not turn off on command.

[Paul Andrews]

One other note, the chip behaved better if I used 3.3V on the inputs rather than 5V. If Greg were right, would pull-down resistors on the inputs help? Not that I plan to test it at this point.

[gregebert]

Ideally, you want a totem-pole driver for the SN74568 because it will shunt-away some of the leakage current that would otherwise get amplified.

You've got me stumped on the 3.3V vs 5V observation. The only thing I can think of is the 3.3V driver has stronger pulldown than the 5V driver (need a scope to confirm).