what's going on here?? (much head banging)

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
OK so I decided to finally get around to matching the transistor pairs in my amp, and perhaps at the same time getting rid of the noise in the right channel (sort of a low buzzing but at multiple freqs can be heard in woofer, mid and tweeter....)

well I matched the trannies for hfe, and chose the pairs with the highest hfe (used same hfe values left and right too), I put some good quality cermet trim pots in for the bias adjustment and thought everything would be sweet.... wrong.....

Right channel distortion went through the roof.... I suspected I had made a dud solder joint, so resoldered everything.... no luck.... measured voltages around the circuit, and almost everything is the same between channels. However a couple of things have me somewhat stumped....

1. The voltage on the right channel between d6 and d7 is 2V the voltage on the anode of d6 is -0.5mV and the voltage on the anode of d7 is -0.5V what I don't get is where the hell is the 2V coming from???? Is it a case of the amp oscillating on this channel?? I did ac measurements too, and get less than 20mV at this point..... I get the readings with no speaker connected, and input shorted.

2. The other thing I'm having difficulty with is the fact that the collector of Q4 in the right channel has -42V on it and the collector of Q4 in the left has -30V on it... all other voltages measured around the circuit are within about .5V of each other (mains voltage keeps varying so getting accurate measurements is difficult :( rail voltage is fluctuating by 1V)....

I replaced q3 with a 30K resistor (collector to emitter) it probably should ideally be about 31K, but I don't think that is my problem. When I first did this I was getting a reading of about -30V on the collector of Q4 in both channels.... previously I had -20V on the left and -38V on the right (with the transistors in place).

3. Actually one other oddity...... the base of q2 on the left channel shows 20mv while the right shows 32mV...... the voltage on the other side of the 1K resistor on the left channel is no different to the voltage at the base of Q2, but in the right channel it is 10mv less. The 47uF nfb caps have been replaced with 220uF, is it possible that the one on the right channel is leaky (it's a Panasonic FC replaced about a year ago)?? I'm guessing that this small difference on the base of Q2 is what is causing Q4 to turn on harder......

Any suggestions greatly appreciated!

Tony.

edit fixed q numbers, I have two different schematics....
 

Attachments

  • amp.jpg
    amp.jpg
    58.1 KB · Views: 771

anatech

Administrator
Paid Member
2004-06-06 8:31 pm
Georgetown, On
Hi wintermute,
1.) Could be meter resistance, need a 'scope to see oscillation as most meters HF responce drops quickly over 200Hz or so.

3.) The base voltage depends on base current and DC resistance to ground on the input side. Tail current divided by the beta of the transistor at that current times the resistance to ground. Very accurate. Your cap sounds as if it is leaky, depending on your meter of course.

2.) That voltage will be variable to some extent. Depends on the current draw to achive circuit balance. There needs to be a normal bias to be meaningful.

Didn't look up your parts so I won't comment.

-Chris
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
ok I'll look into that :)... I have pretty much decided that most if not all of my problems are due to sub standard soldering. I've decided to pull the chasis apart (what I should have done in the first place) to get better access to the pcb, and methodically go through and resolder all of the joints. going to be a bit of a pain but hopefully will pay off.

Tony.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
anatech said:
Hi wintermute,
1.) Could be meter resistance, need a 'scope to see oscillation as most meters HF responce drops quickly over 200Hz or so.

3.) The base voltage depends on base current and DC resistance to ground on the input side. Tail current divided by the beta of the transistor at that current times the resistance to ground. Very accurate. Your cap sounds as if it is leaky, depending on your meter of course.

2.) That voltage will be variable to some extent. Depends on the current draw to achive circuit balance. There needs to be a normal bias to be meaningful.

Didn't look up your parts so I won't comment.

-Chris

Thanks Chris :)

I tryed replacing the nfb caps with some brand new 47uF caps and got the same result.... I guess my meter could be part of the problem, but the weird thing is it gives consistent results channel to channel.

on 3. If I understand you, small differences in the 4.7K resistor on the collect or of q2 (in each channel) could account for the difference in current flow through the transistor and hence the different base voltage? in any case it probably isn't important...

on 2. I actually had the two transistors showing the same voltage at one point, now they aren't, as I said in my previous post, I think that I've done some sub-standard soldering (due to being lazy and not properly pulling the chassis to bits and soldering under non ideal circumstances :rolleyes:

I'll go through and resolder everything properly (ie desolder, clean up leads, clean flux off board, resolder properly) and hopefully I'll have some joy :)

Tony.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Thanks clem_o and edl I'll try the mods :) I haven't got very far with the resoldering yet, but I'm hoping to make a dent in it today :)


edit actually clem_o, with regards to the 47pf cap, do you mean from the bc556 side to ground to shunt any high freq to earth, or actually in parallel with the resistor, so the any high freq's in the output are fed back??

edit: edl I did at some point during my testing get oscillation (evidenced by the coil and resistor in the zobel getting really really hot,) but I think it may have been because I had my input cabling and speaker output cabling in a mess together on the floor (though it is quite possible that the amp is oscillating at a lower level all the time)

If the mods can fix it so it won't oscillate at all then great :)

Tony /probably needs to do some reading on the causes of oscillation ;)
 

clem_o

Member
2005-06-14 6:34 am
Manila
Hi - 47pF across (in parallel with) the feedback resistor, resulting in a gain drop as frequency goes up. You want to reduce the gain at HF so that there's less chance of the amplifier having enough gain (at the wrong phase) to oscillate...

If this helps but doesn't totally solve the problem, you may have to try slowing down your VAS stage with a little bit of local feedback, or look into layout / grounding problems (though the former is improbable since the other channel seems to be ok.. )

Hope this helps!

Cheers
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Thanks Clem_o :) I think I get it now the cap changes the phase by 90 deg so it will tend to cancel out the high freq’s a bit. There are slight differences in the layout from left to right but only pretty minor, the amp is actually an integrated amp with everything on one giant pcb. pic here ---> http://www.diyaudio.com/forums/attachment.php?s=&postid=283801&stamp=1071613264

The one thing I have been wondering for a long time is that the right channel has a relatively long (approx 25cm) jumper wire to get the speaker output across to the output relay and zobel which are on the left hand side of the pcb. I've replaced it with 12 ga wire but it has often bugged me.

One other small detail which I'm going to address, is that the length of wires from the PS to the separate channels is not equal (both channels share the earth return) the left ones are relatively short at about 20-25cm and the right ones are much longer at about 40-45cm) I've noticed in some other posts people saying that length of the PS wiring should be kept as short as possible, so this may be part of my problems too.
 
I like to thinking of the feedback capacitor simply as a frequency-dependent resistor whose 'resistance' goes down as frequency goes up, causing the closed-loop gain to drop (but yes, it does cause the phase of the feedback to change)...

Certainly wire and trace lengths can have an effect on the stability of the amp - though I doubt if changing jumper wires to thicker ones will provide greater stability, since at HF these wires tend to look pretty much the 'same' - an inductance! The point is, the amplifier isn't supposed to be doing "anything" at those frequencies where such wires begin to look inductive - you'd want to slow down the amplifier circuit so that no gain is left to do anything funny in the MHz region... :)

You can try shortening the wires as best as possible, it may have an effect. However, I'd suggest trying to goal for an amplifier that is more than 'marginally' stable. If you're saying that input cable lengths and speaker wire lengths can get the Zobel to heat up, that is definitely an indication of marginal stability, which will tend to affect the sonics, even if you "think" you aren't reaching that region of instability under normal operating conditions...

Cheers!
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Thanks Chris and Clemo_o, the resoldering job is slow but progressing, I'm making sure that I can't blame faulty soldering for anything :)

Found one thing that could have been an issue so far, the pcb pad on the base of q6 was half gone, didn't see until I cleaned up with pcb cleaner... probably the result of some careless desoldering years ago with a less than adequate iron (in a previous attempt to work out why the voltages channel to channel were quite different).

anyway, I'll keep plodding away :)

Tony.
 
Tony - If you can borrow an oscilloscope, I think you'll find the problem a lot easier to pin down... Chris is right, if the power supply wires are adequately bypassed, inductance shouldh't be an issue. Then again, if the bypass capacitors are bad... the easiest way to tell is to have a 'scope handy...

Yep, just keep plodding - it'll be worth the effort!

Cheers
 
Hi Wintermute,

I would like to see emitter resistors in the input diff'l stage of this amp - it will reduce the stage gain, increase it's bandwidth and signal handling and increase the stability of the amplifier. Then grab a CRO and test with 10KHz square waves for peaking and try feedback C as suggested. 47R or 100R should make a difference without slugging the performance.

Cheers,
Greg
 
That's a good idea - emitter degeneration should drop the O/L gain a bit for better stability and linearize the front-end at the same time... That'll increase the overall THD though, but then again there seems to be some consensus now that there's little correlation between THD and what is perceived as 'good sound'... (and maybe people are gonna flame me for this??!!)

Oh btw there was a suggestion of finding alternative, higher voltage transistors to the BC556 - without the emitter degeneration, make sure that the 'new' transistors do not have a much higher beta than the originals, as this change may move the amplifier towards greater instability...


Cheers!
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
Well after many visits from murphy, I have finally re-soldered the entire left and right channels of the power amps.

Good things:

the distortion in the right channel has gone. Thank goodness.

the noise level on the left channel is even better than it was before.

Bad things:

the left channels THD and IMD is a bit worse than it was before (but not detectible at least to me in listening)... went from .03% THD to .09% THD

the noise on the right channel is still there.

The preamp section definitely has a problem..... I get shocking IMD (very noticable even at moderate levels, tweeters are crackling) when playing through the preamp. If I output direct from the sound card into the power amp then no noticable distortion even when cranked right up.

I tried using my pc based scope to check a 10Khz square wave, but either the pc can't generate one properly, or it can't record one :(

I haven't tried the emittor resistors or feedback caps yet, wanted to try and get the square wave measurements before and after....

One of the visits from murphy was that my volume pot carked it and I replaced it, I'm wondering if it could be a source of distortion in the preamp..... I think this amp is possessed!! Time to have another look on ebay for a scope I think.

Oh and one other nasty, the crosstalk has become very poor.... went from about 68db to 33db :( but I suspect that is PS related...

Tony.