Piggyback problems with a PEAK component analyzer (Brett?)

[zjokka]

Almost dare not post on the subject again, Brett treatment was so thorough:

connect the bases,
connect 10k resistors to the emitters and tie them together to make up the new emitter,
connect one collector to the circuit (piggybacked TR) and leave the other open (piggybacking TR)

I had the original thread and update printed out and kept it in the toilet for reading.

Couldn't wait to get started on this because I got my dad's transitor analyzer (first version), so this morning was prepared for some basic breadboarding for testing the transitors. Just plugging one transistors in the breadboard, with three connectins to the analyser, to read normal Hfe and determined the e b an c.

It has always been a why the new model PEAK analyzer can measure leakage, as I tend to believe it measure the Hfe without leakage. Most transistors measure equal on higher in the DMM.

BC438 (Hfe 270) + BC438 (Hfe 270) = 5
BC337 (Hfe 385) + BC337 (Hfe 400) = 40

This all looked promising. I then started mixing types and got quite unoriginal results.

BC337 (Hfe 385) + BC438 (Hfe 270) = 5
BC337 (Hfe 385) + BC3904 (Hfe 185) = 5

I then tried BC538 (Hfe310), BC208 (hfe 200) but always got 5, irrespective on which was piggybacked. I was getting worried here.
The BC337 with Hfe 40 would be interesting, as we would be wanting gains of 70-100, if I could bring in the resistors.

Now, after I attached any resistance on one or both emitters, the piggyback read in my analyzer as a DIODE. After a lot of trouble I managed to attach leads to the construction to read it in my DMM, but no change.

Because of their construction, you can't measure a meaningful hFE of piggybacked transistors in a DMM.  But I've been doing some measuring of hFE of piggybacked transistors in their positions in a fuzzface.  The results surprised me a fair bit.

Does this mean the circuit has to be powered to test gain with resistors? Or is this a problem with my analyser? Or is this just a physical problem?

Also, I checked all the threads, and the Miss Piggy schemo is gone. They were linked to in below thread but as you see all the pics are gone:
http://www.diystompboxes.com/smfforum/index.php?topic=21245.0


thanks
zjok

[zjokka]

ok, right, I figured it out in the end. still hard to wade through some heavy threads though.

using the analyzer allowed me to do some testing with some pot, a pot and some ready piggy's

BC549B and BC270 I picked for its high gain: 405 and 430 respectively
2N5088 as a standard common device: 205
2N4401 as a non-fet personal favorite from my multiface experiments: 270

I'm still wondering:
*would it be better to start with a higher gain transistor and add little resistance instead of the opposite.
*will the sonic qualities of the piggybacked favorites be preserved after PB-ing?
*there were so little experiments with the original FF gain rnage (70-130)
*would adding between 4-6k to a SI tranny always mean jackpot?

will be able to tell once I'm back from the basement.

later,
zj


adding the resistance to the emitters had the following effect on the transistors

[brett]

Wow.
That's some great work.  After word got out about the qualities of the 2N2369A (hFE about 70), I mostly lost interest.

Your results indicate that we should have been using a lot less than 10 kohms for the piggybacking resistors in fuzzfaces.  About 3.3k for Q1 and 6.8k for Q2 look good. I suspect that hFEs in Si fuzzfaces should be a bit higher than the standard 70 and 130, but I haven't built enough to say this for sure.

This is just the excuse that I need to chuck those annoying Ge fuzzfaces, and build a piggybacked version of Joe Davisson's twist on the FF: the Antiquity fuzz.
cheers

PS Sorry about the lost images.  I've got broadband now and changed my ISP.

[RedHouse]

Because of their construction, you can't measure a meaningful hFE of piggybacked transistors in a DMM.  But I've been doing some measuring of hFE of piggybacked transistors in their positions in a fuzzface.  The results surprised me a fair bit.

Refresh us if you will Brett, why can't we measure a piggy back pair in the DMM? I made a little jig that has two transistor sockets and a 47k trimmer for the resistor and it seems to measure a pairs hfe just fine. My DMM tests at a 10uA base current at 3v collector

I know the FF runs at a different base current and collector current but that shouldn't mean a DMM on it's own can't measure a piggy back pair for other purposes (does it?)...hmmmm

Anyway I gave up on trying to calculate the hfe of pairs for the FF and now just use a jig to set things as needed per each FF build I do anymore, I find that dialing it in by hand has much greater success than trying to find a formula to stick to, plus it gives me the chance to dial-in the sound and personalize it, while it's being played through.

Here's my Fuzz Face "jig":


nearly everything is adjustable, caps and transistors are socketed, volume and fuzz pots are on 3-pin jumpers so they can sub'd for testing also. A set of standard input/output caps are on a switch to A-B test changes there etc, etc. It has little brass test point terminals in the appropriate places so when I dial-in the circuit I then shut off the power and yank the transistors and measure the trimmer pots. It even has on-board LM7809 regulator so I can A-B test battery -vs- power supply, and last but not least it has the Pos/Neg ground option that put to rest (for me) that age old debate about Pos Gnd running better (w/o oscillation) than Neg Gnd.
(it doesn't BTW, when one has oscillation it occurs on both)

The two trimmers in the center of the board under the sockets for transistors are for setting the piggyback resistance on a pair.
(I use like pairs, usually 2N5088)

I have noticed that when I set up a pair then remove them and place them into my DMM jig (2 transistor sockets and a 47k trimmer) to measure the resulting hfe it is usually much higher that our traditional target of 70-130, but since it sounds great I've let go the notion of landing on the traditional numbers ...in trade for great sound.

Anyway back to the thought of measuring hfe in-circuit, I don't see why that's important unless one is trying to imitate the Ge setup with Si.
(which won't sound exactly the same anyway)

-Brad

[R.G.]

Your results indicate that we should have been using a lot less than 10 kohms for the piggybacking resistors in fuzzfaces.  About 3.3k for Q1 and 6.8k for Q2 look good. I suspect that hFEs in Si fuzzfaces should be a bit higher than the standard 70 and 130, but I haven't built enough to say this for sure.

I thought that was obvious from the first work.

Remember that when I proposed piggybacking, it was without emitter resistors and we got gains not of half the original like I thought but about five?

I surmised that this is because the internal base-emitter diode in the "active" transistor and its internal Shockley resistor have a higher voltage drop than the "piggyback" transistor, so the piggyback transistor is eating more than half of the available base current, and does so until its forward drop equals that of the active transistor's base.

The resistor between the emitters adds some voltage drop to the piggyback transistor's base-emitter, and offers a way to dial back down the proportion of  base current that the piggyback transistor steals from the active transistor.

As such, we would expect that you would get the lowest effective gain with the inter-emitter resistor equal to zero and the highest effective gain when that resistor is infinity (that is, the piggyback disconnected).

These measurements confirm that story pretty well, exactly as one would expect.

[brett]

Hi
that's a good summary of it all, RG.

RedHouse:

Refresh us if you will Brett, why can't we measure a piggy back pair in the DMM?

As RG noted, the original hFEs were about 5.  However, I used a couple of these in fuzzfaces and they worked ok, so I jumped to the conclusion that the DMM value was wrong.  But of course, it is right for what it is measuring.

To me, the most likely reason why piggybacked transistors can sound like hFE=50, but measure 5, is due to differences in Ib and Ic between the working fuzzface and the DMM test conditions.  I'm not an EE, so there might be lots of other reasons, too.  In conclusion, the DMM results are only an indicator, or comparative tool, with respect to piggybacked transistors, especially for low emitter-emitter resistance and hFE.

PS I'll dig out my original piggybacked FF sometime and double check that it doesn't have emitter resistors, but plays ok.  If the hFE really was 5, it should have an input impedance of about 2k, and totally lack treble.  I'll measure the real small signal gain and the real input impedance while I'm at it. 

cheers

[RedHouse]

...the original hFEs were about 5.  However, I used a couple of these in fuzzfaces and they worked ok, so I jumped to the conclusion that the DMM value was wrong.  But of course, it is right for what it is measuring.

To me, the most likely reason why piggybacked transistors can sound like hFE=50, but measure 5, is due to differences in Ib and Ic between the working fuzzface and the DMM test conditions....

Ahh that sounds right, thanks Brett, I've got some great mileage out of your work on the piggy-back thing.

[R.G.]

It's only a suspicion, but I'll bet that the gain reduction does not remain constant across a range of values of base current.

Since the active base has an effective Shockley resistance and the passive one does not, it would be moderately amazing if the gain reduction was linear.

This is probably a Good Thing, in that it is even more asymmetrical than a transistor all by itself. It may be more Fuzz Face-y than a realy germanium FF.

It's worth some more testing.

[zjokka]

After word got out about the qualities of the 2N2369A (hFE about 70), I mostly lost interest.

Of source there are low gain SI transistors, but when I experimented with them (like BD139) in the SI Fruzz, I like JFET and Mosfets more. So didn't really mind. Furthermore, finding a 120 hfe SI transistor is not obvious.

I suspect that hFEs in Si fuzzfaces should be a bit higher than the standard 70 and 130, but I haven't built enough to say this for sure.

Really is there any truth the gain range myth for fuzz faces? I it's really true that you build a nicely fuzzing one with gains below 50, what can is left. Ok, the historical fact is that the first FF probably had these gain ranges because it made them stable and the AC128 was probably supposed to have a gain somewhere in this range.

Still I have built my piggys to the values 70-75-80-85 etc. Just to try. Types I don't care about...
On the test board now I have a 85 for Q1 and 100 for Q2 and it realy rocks and is very steady, used 2.1K as bias resistor, determined by ear

RedHouse,

that is some nice Fuzz Lab you have there. The way to go if you really want to get inside a circuit and get the most of it..
Is it an etched layout?


Here some pictures of PB transistors:



Once the transistor's soldered into place, I remove the plastic straps

zj

[RedHouse]

RedHouse,

that is some nice Fuzz Lab you have there. The way to go if you really want to get inside a circuit and get the most of it..
Is it an etched layout?

Hey great name for it "Fuzz Lab" I'll keep that if you don't mind.
(way better than "FF Jig")

Yes it's an etched board, it's some real cheapass board material I use for prototyping, not decent FR. I had a smaller perf-board jig I used in the past that had a few option on it, and I built and sold a several dozen FF's dialed-in with it, then a long time passed when I didn't build any for a while (couple years) and when the need arose again I decided that it was time to get this tweak-mode thing out of the way forever and built this all encompasing board that had everything (but the 100k feedback resistor) adjustable.

I also made other "jigs" or testbed units as it were, here is my Univibe testing machine, has all adjustable phasing caps and on-board FF and Octavia (you can barely see the double (Brett style) transistors in the lower left corner of the PCB) and again built from that cheapass PCB material:
(not sure what it is, seems like pressed pulp material, I got them at Maplin's in the UK when I was over there last)



-Brad

PS; here's a short essay on sheilding a Strat:  http://users.isp.com/brad_anne/guitar/shield.html