fet boy

[jarlehal]

I'm not 100% sure my bd139s are placed the right way. All mine are facing the same way. I used the small line on the pcb and on the transistors as guidance, and got those aligned. Maybe that was wrong. Think I got all the resistors in the right place. I'll check this later today. I could try to get some pictures up for you, could help some other kids someday too

I measured the 139s with a cheap DMM, not sure if it's accurate enough. I'll have a rerun tonight, on the ones not mounted yet.

 

[Ptownkid]

If they measure out that hugh, I'll get you a pair that matches the correct Hfe

 

[Samuel Groner]

You definitely need to get closer to 20,000 than that.

20 or 20000? A BD139-16 will not reach any of the two figures--check the datasheet...

Why in the world do people spend days with selecting transistors instead of just changing the design to a reasonable current source..? :?

If I get time I'll draw a suggestion for an improved design, addressing the transistor selecting, heat and stability issues while (presumably) leaving the sound essentially untouched.

Samuel

 

[Ptownkid]

it's 20,000, and it's achieved like this...

"One of the biggest hurdles was to overcome the fact that 'original' MPSA14 Darlington has an hFE of 20,000! The away around this was to use a pair of either BC550A or BD139's transistors that when, multiplied together would bring you close to the original value."

taken from

http://www.diyfactory.com/projects/fetboy/fetboy.htm

 

[Samuel Groner]

I thought you were talking about Q4/Q8--that's the transistor which hFE has an influence on quiescent current.

Forget about selecting Q3/Q7, its hFE has no influence on the performance as long as it is high enough. What the statement on the web page you cited want's to say is that the combined hFE needs to be at least 20000 and not exactly 20000; the hFE of the MPSA14 isn't exactly 20000 either--it is at least 20000...

If you have a BC550C for Q2/Q6 and a BD139-16 for Q3/Q7 there is no way the combined hFE could get below 40000. So once more: just stuff the transistors in.

Samuel

 

[Gus]

Samuel

People have suggested a "better" design With a CC etc. IIRC In one of the threads someone posted S.H. liked the "sound" of the circuit as drawn.

For the "stock" build

IMO If people want to build this they need to share the specs for the FET that "bias up" IDSS and Vgs off might be a big help and the hfe the transistor.

 

[Samuel Groner]

I remember that thread but I have some more ideas.

My main point is that there is no need to select any of the BJT transistors for hFE--at least as long as you use the trimmer shown in the schematic to set quiescent current of the output follower.

Samuel

 

[insideout]

[quote author="jarlehal"]I've joined the club of struggling fetboys. Got a couple of kits from Bryan and I'm putting one together now. Have a few questions..

The kits contain BD139-16 but I measured hfe, and the lowest one was 182. Most around 185. Do I have to get close to 20000 for the darlington pairs (mul) or will my pair of 185s do? (34225)

Got 20 of the 2n5457, but none of them got me to the 12v ballpark.. all resulted in 15v and above. Have to get a bunch of those I suppose.?.

When I have signal thru it, I have to turn the pot fully CW area to get any output. And there's not much gain. Any suggestions? Is this a result of the FETs dont match? I only feed them 24v, no 48v yet.[/quote]

I'm just finishing mine off, I found exactly the same thing re. the hfe's and voltages. I just paired off the lowest hfe BD139-16's to get the lowest overall hfe, and put them in. So they are wildly higher than 20,000, but no problem trimming to get the 4v across r5 and r10.

I used BD135's for q4 and q8 and put heatsinks on them.

All of my 2n5457's were above 15v, most of them well above. I got the impression I could go through hundreds and still not find any that result in 12v. So I have used the lowest ones and put 5k trimmers in place of r3 and r9 to bring it down to 12v.

<edit for mistake in post> r3 and r8, the 1k82 resistors on the source of the 2n5457's, are the ones I replaced with a 5k trimmer <end edit>

haven't tried any audio yet.....!

 

[jarlehal]

So, I shouldn't worry to much about the high hFE.. That's good.
Got some pictures of the board up here.

Anyone see anything wrong? A little difficult light, but I think you can see those markings on the transistors..

Yeah, I can trim the voltages down to 4V (R5 & R10) too without the 2n5457s in. Haven't tried to get down from 15V to 12V with them in, as that seemed a long way down. Could always try From Matt's info, I assumed you had to get closer to be able to tune it in.

But then again, would FETs that give around 15V before the caps give me this lack of gain and weird pot behavior? Or do I have to start double checking the resistors and stuff (while waiting for a pack of 2n5457s). I measured all the resistors before soldering.

insideout: you mean trimmers on R4 and R9 right? And you got them down to 12v? What's the voltage over R5 and R10 then (FETs out) ? Please report back when you have done your "signal-thru".

 

[Ptownkid]

I hadn't thought of that before, but increasing the trimmer size might do the trick...I'm going to look more into that.

 

[insideout]

[quote author="jarlehal"]
insideout: you mean trimmers on R4 and R9 right? And you got them down to 12v? What's the voltage over R5 and R10 then (FETs out) ? Please report back when you have done your "signal-thru".[/quote]

sorry, my mistake, wrote that in a hurry before going to work!
r3 and r8, the 1k82 resistors on the source of the 2n5457's, are the ones I replaced with a 5k trimmer. I tweaked those to give me the 12v at c2 and c4.

that was done _after_ using the 100k bias trimpots (replacing r4 and r9) to give me the 4v across r5 and r10. I didn't touch those after setting the 4v and moving on to getting the 12v. I'll remeasure those for you as soon as I can (I have no idea whether its important!)
- Mike

 

[shabtek]

futzing w/ 1 of these on perf right now, having similar problems. added a resistor before the trimmer to effectively put 80kOhm (or less--trimmable) where you usually have 47k || 100k trim.
It made for a lower V @ darlington emmiter/ bd135 collector but as that V drops overall I goes up far beyond 40mA...V is being seriously dropped across the 30 Ohmers...not sure if I'll get around to tweaking R3/R8 wonder how that will work...
I recall reading when I started this long ago that one could trim out differences in the FETs.?

edit: 5k trimmer for r3/r8 does not do what we need.
put 20k trimmer for r1...not able to dial in specified voltages

for best sine response (most headroom) running each stage @ 80 mA!

 

[jarlehal]

Took another round with the FETs just now, I found one that got me just below 15v and one just a bit over. With those in, the unit passes signal, the pot work normally and with loads of gain!

So if those trimmers on r3/r8 won't do the trick, I think I just have to order a bunch of 2N5457. The ones I got are Fairchild I think. Might try another manufacturer if I can find one.

Sorry if this is stupid, but does it have to be 2N5457, or are there other FETs that could work aswell?

 

[skipwave]

Try the OnSemi brand 2N5457. Others have reported getting more matches with those (as opposed to Fairchild), as they typically have lower hfe. With Fairchilds I found maybe 1 in 50 that met spec.

This design only meets the spec'ed current draw with all devices selected for hfe. The larger value trimmer to the base of the current source allows devices that are above the spec'd hfe to work, but does increase current draw. Scott reported on his website that he added a series resistor with the trimmer. Maybe changed some part values, too. The link is somewhere in this thread.... I think.

 

[3nity]

Hey we all seem to have troubles getting this suckers working huh?
Mine dont pass audio at all...al i hear its a hum...and i have R10 burned...

Seems like the first stage is a non working but i cant really figure it out...
Any pointers?

 

[skipwave]

Break the design into its separate stages.

You can adjust the values of the drain and source resistors to work with the FETs you have. Chrissugar built an army of these things and he chose a better operating point for the parts he had.
To that end, you could whip together a perfboard testing jig with a socket for the FET and trimmers in place of those resistors, then:

> insert FET
> trim until quiescent current (metering 12VDC at drain pin) and gain are as designed
> remove FET
> measure trimmers and find nearest standard resistor values

The output stage DC operating point is set by the first stage, because they are direct coupled.

Actually, I'm surprised that mpsa14 is considered hard to find. I found lots.

If any of you guys who didn't get any BD139s that work in the darlington pair, I'll send you some mpsa14 to replace them. Just PM me.

 

[3nity]

Guys i have 1 board working and the other not....now is it safe to run the boards with out the Fets just to measure voltages?

Thanks again.

 

[jarlehal]

Yes, I think that is OK. Atleast it worked with my board..

From Matt's page:

1. Power up the pre without the Jfets (Q1/Q5)
2. Measure the V across R5
3. Adjust bias to have 4V across it (the trimmer)
4. Now put a Jfet in (Q1)
5. Measure the V at the node before C2
6. Try all the Jfets you have and use the one that give you the nearest 12V (Most of mine were around 14-15V)
7. If you get 12.3 or 11.7 just adjust bias to have 12V
8. Do the same for Q5 checking the V across R10 and the Node of C4

Ordered some more 2N5457 today. Couldn't find any OnSemi @ my usual places, but I found some Motorolas. Hopefully I'll get some FETs that are useable.

 

[3nity]

definetily something is wrong....i must have a short...cuz the psu gets very hot when the nonworking channel is wired....the other one works like a charm...and both boards have the same values...??

 

[skipwave]

[quote author="jarlehal"]Ordered some more 2N5457 today. Couldn't find any OnSemi @ my usual places, but I found some Motorolas. Hopefully I'll get some FETs that are useable.[/quote]

That's cool. I heard OnSemi is the semiconductor branch of Motorola. I don't think they make little silicon blobs with the old "M" logo any longer.

For you guys struggling with this circuit, it is absolutely worth the trouble. Just remember that the transformer and coupling capacitors determine the sound to a large extent. If you can, try some different ones.

Maybe I'll try out one of the improved designs, with a more solid current source implementation. See if that affects the sound. It would certainly improve reliability and repeatability.