Peak Atlas DCA55 not good for vintage silicon

[LucifersTrip]

It's sad...After all the glowing reviews & recommendations, and a cost of $60-$70, I really expected better.

I just purchased it to analyze a bunch of vintage metal can silicons & germs, and the massive FAIL was with PNP silicon. For the last 30 minutes I went through a bunch (with hFe 15+) and the DCA55 reported mainly "darlington", some "common diode network", a few "digital transistor", and one "germanium".

These are the most recent failures: 2N3640, 2N2184, 2N2182, 2N493B, 2N2274, 2N2183, RCA34594, 2N2185, 2N2646, 2N2187

Regarding gains...it measures in between my Metex 3800 and R.G.'s method for medium gain transistors. It's approx 10% higher than the Metex and R.G. is approx 10% higher than the Peak.
(ie, 2N1132 - Metex:95, Peak:102, R.G.:114 ~~ 2N2303 - Metex: 111, Peak: 120, R.G. 128

For lower gain ones, the Metex is still the lowest, but Peak and R.G. could go either way
(ie: HEP51 - Metex: 41, Peak: 49, R.G.: 45)

The Peak is much more accurate identifying germanium...Very few failures. The gains were pretty close to using R.G.'s method. It could go either way by approx 5% - 15%
(ie: 2N404 - Peak: 207, R.G.: 240, Peak: 101, R.G.: 98 ~~~ 2N43A - Peak: 80, R.G.: 85, Peak: 68, R.G.: 56

Out of all these, I have no idea which is the most accurate...and I'm not sure how the Peak accounts for leakage.

[R.G.]

Out of all these, I have no idea which is the most accurate...and I'm not sure how the Peak accounts for leakage.

Aye, there's the rub.   

A packaged semi tester simply has to make assumptions about leakage and how to measure it. If you have a specific application in mind, you have to know what those assumptions are so you can work around them. And I'm guessing the makers didn't document the assumptions for their buyers.

As we used to say in the 1960s,  bummer, dude.

[LucifersTrip]

Yes, the bottom line is...my ear is the best tool.

[Jez Siddons]

Hi all,

Jez Siddons from Peak Electronic Design Ltd here, the designer and maker of the DCA55.

Apologies for the delay in replying to this post, I've only just found it, otherwise I would have replied a long time ago.

I appreciate the points made by the OP and also the replies.

I'll try and cover some of the points made, forgive me if I miss anything.

1. The DCA55 booklet is fairly detailed and shows the test conditions used for measuring gain. For example, the DCA55 measures gain for a collector test current of nominally 2.5mA and a collector voltage of around 2-3V. It is common for many other gain testers to measure at different currents, some don't even specify the collector test current (I know my multimeter doesn't specify the test current at all). Of course, gain can vary at different test conditions, so that can explain why measurements from one piece of test gear to another can be different.

2. The DCA55 does take into account leakage current when measuring gain. It measures the leakage current and uses that measurement to see what *change* of collector current has occurred as a result of it applying some base current. That change of collector current divided by the base current then gives the true gain, regardless of the leakage current that you're getting.

3. Leakage current (and indeed gain) is very dependent on temperature. Germaniums in particular can show a huge change of leakage current with just a tiny change in temperature, typically the leakage current doubles for an increase of just 5 degrees C (8 degrees F).

4. The DCA55 will decide if the transistor under test is a darlington type if the base-emitter voltage is typically larger than 1.0V for a base test current of around 4mA. Additionally, if the transistor has built-in resistor shunts across the base-emitter juunction then the threshold is dropped to 0.80V typically. This means that it is possible that a silicon transistor can "look" like a darlington to the DCA55 if it's base-emitter voltage drop is a little larger than many modern silicon transistors. This is particularly true for vintage silicon transistors as their base-emitter junction is very ohmic (quite a soft knee). The pinout, the gain and the leakage current however are all still valid and the "darlington" message doesn't affect any of the measurements. All these thresholds are explained in a little more detail in the user guide.

5. If you're seeing some transistors come up with unexpected messages on the DCA55 then it is possible that the part being tested is faulty (or a little poor). For example, if the DCA55 comes up with the message "Digital Transistor", that can happen if there is large resistance in the base line (digital transistors are transistors with built-in series base resistors). Another example of an odd message is the "Diode network" message. That can happen if the DCA55 simply hasn't been able to get the transistor to exhibit a gain of more than 4 (4 is the minimum acceptable HFE for the DCA55). In that circumstance, the DCA55 falls back to looking for diode (PN) junctions. It does that because you can get diode network components (typically in SOT23 surface mount packages) and the DCA55 needs to cope with those too. These are all explained in more detail in the supplied user guide.

By the way, the current user guide is available here: http://www.peakelec.co.uk/resources/dca55_userguide_en.pdf

The user guide also mentions that a full refund is available to anyone who isn't happy with their instrument in the first 14 days after purchase for any reason at all. Additionally, all our products have a 24 month warranty.

Also, if anyone has any concerns about whether their unit is behaving properly then please do get in touch. We are a flexible company and we can investigate problems with a view to making improvements where possible. Sometimes this has been done within just a few days of a customer mentioning an unusual part or a special request.

I hope that covers most of the points, if I've missed anything then please let me know.

Regards,
Jez

[digi2t]

Does the version number make a difference? Mine is 2.8. You see the version when you first turn it on.

Maybe the older version firmwares are less accurate? I had a bit of trepidation before buying mine, because I read that many people had the same complaint as Luce. In any case, I took the plunge, making sure that I got a unit with the latest firmware.

Never had a problem with mine. Old trannies, new trannies, jfets alike. The only thing it doesn't like is something like those Russian transistors with the resistor built into them.

BTW, I read that the firmware is up to 2.93 now. What new in that?

[Jez Siddons]

Hi digi2t,

There are a few minor things that get tweaked with each release of firmware. The most notable differences are here:

V2.0 Added transistor leakage display
V2.3 Improved measurement jitter rejection.
V2.6 Implemented improved power management to increase battery life.
V2.8 Added digital transistor detection. Improved factory tests.
V2.9 Increased acceptable leakage for Ge transistors from 1.25mA to 2.0mA.
V2.93 I/O adjustment for new PCB, no core changes.

Generally speaking, when I've come across any unexpected results I've found that the part does indeed exhibit some unusual behaviours when tested manually on a breadboard, it's unusual for the DCA55 to give misleading results for no apparent reason. We will however always investigate things if any customer lets us know of an odd situation, either to identify what is unusual about a part or to make adjustments to the firmware.

Cheers,

Jez

[PRR]

Thanks for the follow-up, Jez.

> collector voltage of around 2-3V
> base-emitter voltage drop is a little larger than many modern silicon transistors

IIRC, the really old Silicon didn't always perk-up at just 2V Collector-emitter. (Theory says a few hundred mV is all the same until Early appears, but some old-old Silicon didn't believe in theory.) Test-specs from the old days show 5V to 30V. The DIY transistor tester published by GE around 1964 uses 6V (or 6.2V regulated) through 330 ohms.

Obviously you would not apply 30V to some modern super-delicate parts! Even 6V is pretty juicy for unknown parts which *might* be 3V cellphone confections.

I think the DCA55 is flippin wonderful for general use on 99.9% of all active parts.

Just that LT (and others here) are into the 0.1% parts, seeking musical mojo.

LT is sharp enough to roll his own tester, using the actual voltages and currents found in the circuits *he* is interested in.

[waltk]

I recently got one (DCA-55), and I really like it.  There have been a few times that I measured old transistors with less-than-shiny leads, and got flaky or inconsistent readings until I cleaned them.  Not complaining - just mentioning that it's something for the OP to be aware of.

Jez, if I could add something to the wish list - is there any chance you could make it report more information about JFETs?  Sure would be nice to have Vp.  Is there some technical reason this can't be done?

Also, suppose you were able to add this feature - is there any way I can upgrade my own firmware (many of us could burn our own ROMs if the code was made available)?

-Walt

[LucifersTrip]

I recently got one (DCA-55), and I really like it.  There have been a few times that I measured old transistors with less-than-shiny leads, and got flaky or inconsistent readings until I cleaned them.  Not complaining - just mentioning that it's something for the OP to be aware of.

The leads were cleaned...if the transistor is working and the Peak does not make contact with one  of the 3 leads, it'll show "diode or diode junction". If the transistor is working (or not) and the Peak does not make contact with > one lead, it'll show "no component connected".

The majority of misreads I was getting were "darlington", "common diode network",  "digital transistor",  and less commonly, "germanium".

The ones I measured were generally rare late 50's and early to mid 60's PNP silicon. Numerous were "bullet" types. see first post.

I don't expect most DIY-ers to have many of those....as you can see, of the 10 I mentioned above, you can only find 1 on eBay, the 2N2646...which I now realize should not be in my "misread" list since it's a unijunction transistor, which I don't believe the Peak recognizes.
http://www.ebay.com/sch/Electrical-Test-Equipment-/92074/i.html?_from=R40&_nkw=%282N3640%2C+2N2184%2C+2N2182%2C+2N493B%2C+2N2274%2C+2N2183%2C+RCA34594%2C+2N2185%2C+2N2646%2C+2N2187%29

[Pyr0]

I don't expect most DIY-ers to have many of those....as you can see, of the 10 I mentioned above, you can only find 1 on eBay, the 2N2646...which I now realize should not be in my "misread" list since it's a unijunction transistor, which I don't believe the Peak recognizes.

Build a Vox Repeat Percussion with it 

[Jez Siddons]

I recently got one (DCA-55), and I really like it.  There have been a few times that I measured old transistors with less-than-shiny leads, and got flaky or inconsistent readings until I cleaned them.  Not complaining - just mentioning that it's something for the OP to be aware of.
Jez, if I could add something to the wish list - is there any chance you could make it report more information about JFETs?  Sure would be nice to have Vp.  Is there some technical reason this can't be done?
Also, suppose you were able to add this feature - is there any way I can upgrade my own firmware (many of us could burn our own ROMs if the code was made available)?
-Walt

Hi Walt,

In answer to your JFET question. Yes I agree, it would be great to display more information when testing JFETs. The reason it's tricky for the DCA55 is that its limited in the range of voltages that it can generate. The voltages generated are currently limited to 0-5V (with 10bit resolution) and that span isn't enough to reach Vp for many JFETs. We're currently looking at this to explore ways of deriving pinch-off (and other parameters) using the limited range of voltages that we have available. Like I say, it's tricky but worth investigating. We may or may not succeed, especially in light of the fact that the ROM is up to the ceiling with code already.

In terms of performing the firmware upgrade process, sadly it's something that has to be done here. I agree that many folk have their own programming kit and that could be used but we prefer to carry out the process here. That's because there are a few boards revisions we need to take into account and there are 3 different micros used too, all with their own sub-versions. What's more, we like to do a full test here after any programming op to make sure the whole thing is working to within nice tight limits.

For instruments still in warranty, we perform any upgrades for free. After that period, we only charge £12.77+VAT plus postage. That includes the full test, new battery and new user guide too.

Cheers,
Jez

[Jez Siddons]

I don't expect most DIY-ers to have many of those....as you can see, of the 10 I mentioned above, you can only find 1 on eBay, the 2N2646...which I now realize should not be in my "misread" list since it's a unijunction transistor, which I don't believe the Peak recognizes.

You are correct, the DCA55 doesn't support unijunction transistors. We shall investigate that though, they're very rare but interesting devices and still fun to play with.

Cheers,
Jez

[waltk]

it would be great to display more information when testing JFETs. The reason it's tricky for the DCA55 is that its limited in the range of voltages that it can generate. The voltages generated are currently limited to 0-5V (with 10bit resolution) and that span isn't enough to reach Vp for many JFETs

Ahhh, I see.  Well, for most of the JFETs I use (e.g. J201 2N5457, MPF102), Vp is generally less than 5V.  Vp varies a lot. but even if the range of V_GS(off) is -.3 - -8V for a part, it would be useful to know the values that do fall within the voltage limits of the DCA55.  Perhaps you could just indicate when it is out of range.

LT - sorry to hijack your thread with OT requests.  Sounds like you have an interesting collection of old parts.

[LucifersTrip]

I don't expect most DIY-ers to have many of those....as you can see, of the 10 I mentioned above, you can only find 1 on eBay, the 2N2646...which I now realize should not be in my "misread" list since it's a unijunction transistor, which I don't believe the Peak recognizes.

You are correct, the DCA55 doesn't support unijunction transistors. We shall investigate that though, they're very rare but interesting devices and still fun to play with.

Cheers,
Jez

That would be great...but, I would not even need the Peak to give me any info at all regarding the Unijunction transistor....just some indication that it may be one and is not a "common diode network" or "led or diode junction", which it tells me now.

For me, a big problem is that when I get that "common diode network" or "led or diode junction" message I have no idea if I have a bad BJT or good UJT.

I've taken the time to sort and look up in guides about 75 - 100 UJT's, but I still have this "small" selection (and many more) remaining (many have no markings on the case or have un-findable house numbers).

As I wrote in another thread, I have that bag labeled "DOA" because of high leakage, super low gains, one leg broken...but the most common reason I placed them in that bag was because of "common diode network" or "led or diode junction"....but are they really DOA?....or UJT.

[PRR]

Approached in ignorance, a UniJunction should be a "crummy diode". At low voltages it is a diode-drop plus several K of resistance to either end. Mostly you'd assume it was a diode or twin-diode with nearly-failed contacts. And with several K in series, a "crummy diode" isn't much use... toss it! But a 3-leg crummy diode *might* be worth flagging as "possible unijunction".

With a clue, anybody who would have use for a uniJ can whack-up the basic oscillator and get actual results.

For more fun: there's Shockley's original unijunction and the modern PUT. They overlap but are not the same.

[LucifersTrip]

With a clue, anybody who would have use for a uniJ can whack-up the basic oscillator and get actual results.

Could I plug a UJT into this circuit to test if it "works"?  Or will that not give me a complete answer, since it could "work" to varying degrees?


btw, is there a quick alteration for 9V supply...

[PRR]

> is there a quick alteration for 9V supply...

With PhotoShop: cut the "6", turn 180 degrees, paste the "9".

And speaker power rating must rise from >0.05W to >0.1W.

> could "work" to varying degrees?

UT circuits vary so much, that you'd best test in the actual circuit.

[LucifersTrip]

> is there a quick alteration for 9V supply...

With PhotoShop: cut the "6", turn 180 degrees, paste the "9".

you got me on that one..lol

UT circuits vary so much, that you'd best test in the actual circuit.

sadly, I figured that might be the answer. thanx again

[buildafriend]

Hey guys,

I recently had a similar issue with the DAC55. It really is a genius piece of engineering, unfortunately it does not seem to be getting me where I need to be with lower gain PNP transistors. A forum member redirected me here after I posted asking for help with transistor testing.

http://www.diystompboxes.com/smfforum/index.php?topic=101965.0

I know nothing more about unijunction transistors and programmable unijunction transistors than their schematic symbols. I did get some short schooling on them but all that I really took with me was enough to try to get by in a job interview. I wouldn't mind playing with some just to gain some more skills.