♩ I'm going slightly mad... ♩ Transistor Craziness!

[Ti Pwun]

OK, I know I've been pretty ill recently so I haven't been able to work on some of my projects, but I'm now back on the repair of a recentlty acquired Goodmans Module 80 tuner/amp.

I had already determined that the output transistors needed to be replaced, so I ordered 4 x MJE 2955.

But now that I have the old ones removed, I am confused.

According to the silk screen on the PCB, the pin order from left to right when looking at the transistors with the bent pins towards you (see images below) should be EBC.

The new transistors, however, are BCE. I simply ordered 4 x MJE2955 and never gave it another thought; that's what was in there and it was going to be a quick out and replace job.

Now onto the testing of the old ones. I marked them with one, two, three and four dots for the purpose of taking the following images. Each channel on the schematic has two of these PNP transistors marked TR21 and TR22.

I have labelled my images TR21a, TR21b, TR22a and TR22b. The three leads on my tester are connected to the transistor pins from left to right in the order R, G and B in each case.

As you can see from the pics below, here is what I get:

TR21a - CEB
TR21b - ECB
TR22a - ECB
TR22b - CEB

None match the silk screen (EBC) shown on the PCB and my order of the transistors may not be the same as they were when installed. In any case, they should all be the same!

My first thought was that there must be a fault with my tester. I replaced the battery but it's the same. I also tested around 20 new transistors of varying types and get no errors on my tester.

Am I missing something? I am not in the best of health right now, I know, but with two of them reading CEB, the other two reading ECB, none of them matching the EBC on the silk screen and with the new transistor showing BCE, not the expected EBC as per the silk screen, I am wondering what I'm missing.

Any input welcome.

[Red to black]

I am not sure I follow you ? if the old transistors needed replacing then surely they must be faulty no ?
If they are indeed faulty then why would you expect the transistor tester to be able to identify the leads correctly ? at best (assuming the old devices are faulty) then it would/could confuse that type of tester ?

I looked quickly on line at the data sheet for type number MJE2955 and the first few results came back as a different package to yours in the picture, namely they show a T220 package where the pins L to R from the front are indeed BCE.

incidentally the first two datasheets I hit upon had suffixes of T and G to the base number, what suffix ? if any, do the original type have, as it is not clear from your picture

[julie_m]

The silk screen probably is wrong. Power transistors almost always have the collector in the middle. (You can actually fit a TO220 device in place of a TO3, if you are feeling sufficiently bodgy, just by bending the outside leads and cutting off the middle one.)

And a transistor tester can easily get confused by low-gain devices. A transistor will actually amplify with the collector and emitter reversed; just not as well. This is how devices like the Peak Atlas determine which lead is which: by trying it both ways and assuming the way that gives more gain is the right way. A small-signal transistor like a BC547 will have a gain of a few hundred the right way around and barely 20 with C-E reversed. A power transistor will have a lower gain anyway, and more so at low test current; it can occasionally mis-identify the collector and emitter.

[Red to black]

Hi Julie,

I knew about the possibility of reduced gain on power transistor connected backwards as it were on that type of tester, it baffled me why testing possibly/probably defective parts would make any sense to that type of tester at all.

The OPs original devices also look to be TO-126 type casings? or the larger version ? which TO-number escapes me for the moment* where by looking from the front they are ECB from L-R, or the opposite way to TO-220 types, like you say I don't recall any of these types of power device having the Base connection in the centre.

Edit: * possibly TO-247 ?

[Radio Wrangler]

I definitely don't trust that silk screen marking.

The way those transistors are built has the collector (=substrate) bonded onto the heatsink face metal, which is formed also into the centre lead. So the centre pin is always the collector unless someone has tried really hard and spent money doing it.

If the screen marking is tight, then the transistors need their little legs insulating and twizzling round.

David

[julie_m]

Well, output transistors don't always fail short between all 3 leads (though that's usually the easiest fix; eight times out of ten, when this happens, the failure of the output devices will have blown the fuse before anything else had time to get damaged)... If the fault is just that the gain has gone low, that would explain the readings.

What did you measure for hFE on those transistors?

[Red to black]

Well, output transistors don't always fail short between all 3 leads...

Agreed, what I meant was if they had failed short it would be impossible to identify which lead was which by solely testing between leads at all, no matter which tester or method used, if they have gone O/C then the same applies, and if leaky then at best dubious results, low gain or noisy you might be able to identify depending on how and what tester was used.

I had one of those Peak transistor tester/analysers at one time, and while they are very good most of the time on new parts, as a repair man I found they were frequently easily confused with failed components, I actually swapped mine for the Peak ESR meter as I found that much more useful in a repair workshop

[TrevorG3VLF]

It seems strange that the transistors are all the same. I would think that it is more likely that each pnp is paired with an npn.

npn are more popular, isn't the 2955 a pnp?
What are the numbers on the original transistors?

[Andrew B]

I usually found the 2955 PNP / 3055 NPN in combination with each other and always BCE with the leads towards me and the heatsink side down

- - -
B C E

Andrew

[Red to black]

Originally Posted by TrevorG3VLF

It seems strange that the transistors are all the same. I would think that it is more likely that each pnp is paired with an npn.

npn are more popular, isn't the 2955 a pnp?
What are the numbers on the original transistors?

The Goodmans Module 90 I believe used split rails (+ and -) and used complementary NPN/PNP output pairs MJE3055 and MJE2955 respectively, the OP is however referring to a Module 80 which used only a single rail +ve Gnd, hence the pair of PNP MJE2955 output transistors per channel.
I will however stand to be corrected

Edit: found a scrappy diagram, attached a snippet of the output for one channel.

[Radio Wrangler]

Designed by someone who didn't quite make the changeover from Germanium?

David

[Red to black]

Yes David,
It does seem to be based on an older earlier design, +ve Gnd and all, these were early or mid 70s I think, yet the design seems almost 60s

[dseymo1]

Surely the best way to be sure is to trace the circuit?
A few quick checks of the tracks should confirm what goes where.
If necessary, the suppliers should be able to verify the pinout of the new transistors.

[Ti Pwun]

I've abandoned the schematic for this, at least for now. TR20 is marked as a BC301, which is PNP, but on the drawing it shows an NPN symbol.

The original transistors are all MJE2955 but I was following someone's comment elsewhere that '...the Module 80 uses MJE2955 output transistors'. I think this has led to some confusion and assumptions. Given that I've now found definite errors on the schematic, I need to dig deeper.

One of the problems in terms of reading the order is that the centre pin was connected to a short lead that went off to the C connector. So ECB is in fact correct. And that makes sense. I removed the wire before I became ill and only just remembered because of the two short I found on my tub of parts.

Two of the originals do show that so the two that don't are going to be the faulty ones.

Julie's is the answer here. Hfe is as follows:

TR21a - 16 -CEB
TR21b - 27 -ECB
TR22a - 40 -ECB
TR22b - 7 - CEB

Low Hfe on TR21a and TR22b explains the EC misidentification. Something I didn't realise my tester would struggle with but I'll know in future.

I'm just going to refit the originals. The new ones are BCE (and metal body) and would involve adding wires and possible mounting issues because of the heat sinks being electrically connected to the chassis (no metal on the originals around the mounting hole).

I had lifted two transistor connections originally and determined with my tester that there must be a fault, but the misidentification of the pins shows that to be a bogus test.

Thanks Julie, and all other commenters. It all makes perfect sense now and what I've really discovered is that the MJE2955s I bought are likely to be fakes...

...and that I need to make better notes as I become ill; I completely forgot about the two collector wires I had removed a month ago.

Oh well, I had already replaced all the caps so maybe putting back the original transistors will bring this back to life.

[Ti Pwun]

You're right Dseymo1 - I was planning to do just that until I read Julie's comment. I've never used this tester on very low Hfe transistors before but it all fell into place when she wrote that.

[Ti Pwun]

Redtoblack,

Yes, that's what I'm working from. I had expected an NPN/PNP pair but that's not the case.

Three problems came together to confuse me (my tester making me assume a fault on two transistors, me forgetting that I had removed two collector wires (per channel) prior to my diabetes rearing its ugly head, leading to me reading the silk screen literally and my replacement MJE2955s reading backwards to the originals.

I couldn't quite work out what was going on. But Julie's post put me back on the right path.

Thanks again everyone - much appreciated.

[Red to black]

Hi Ti Pwun,

I've abandoned the schematic for this, at least for now. TR20 is marked as a BC301, which is PNP, but on the drawing it shows an NPN symbol.

A quick datasheet search indicates that a BC301 is indeed an NPN see attached, that actually ties up with the diagram as its C is shown tied to Gnd, which is +ve as already said.
I hope I am not adding any more confusion here

[Ti Pwun]

Thanks, yes it is - my mistake for Googling BC301 and misreading something I saw without pulling a datasheet. A couple more days in bed may be in order!

[Red to black]

We all make mistakes, I hope you are feeling better and back to your old self soon

Ps. I don't think the transistors you bought are fakes, they are BCE L to R looking from the front with the heat sink face down, they are just a different package to your originals.

[Ti Pwun]

The originals are ECB from left to right with the pins bent towards me and the heat sink face at the back but the replacements are BCE when bent in the same way and the heat sink face at the back, so I was assuming any MJE2955 transistor would read the same way, making me think that perhaps the new ones are not actual MJE2955s. But maybe they are.

In any event, the originals have been remounted with new Sil-Pads and I'm about to solder them back in. I won't test it for a couple of days, though - I clearly am not as rested as I thought I was. Hard to keep away from electronics for long, though!

Thanks for the well-wishes. Appreciated.