Feed the amp a 1KHz square wave and 'scope the vas stage. Actually, you can look at the waveforms throughout the amp. In a marginally stable amp, you may see ringing or low oscillation on the waveform. You should have a load on the amp and you can use a low level. Watch you don't burn out the zobel network if so equipted.
-Chris
-Chris
Hi Tom,
Do you use a small value 'C' in parallel with the NFB sensing resistor from the output; typically about 22pF for 10k.
This can reduce 'crispness' but the value should really be determined on an individual amp. circuit basis, either trying a handful of different small value components, or by using an insulated air spaced SW variable and then externally measuring it for your chosen preference.
Cheers ......... Graham.
Do you use a small value 'C' in parallel with the NFB sensing resistor from the output; typically about 22pF for 10k.
This can reduce 'crispness' but the value should really be determined on an individual amp. circuit basis, either trying a handful of different small value components, or by using an insulated air spaced SW variable and then externally measuring it for your chosen preference.
Cheers ......... Graham.
Hi Tom,
Go back, back into the amp .... 'Scope the vas and other stages to see what's going on inside. You can then tell if you tend to overdrive the vas if you turn it up some into a dummy load. I'm sure you don't feel like listening to tones up high. That & you'd take out mids and tweeters doing that.
-Chris
Go back, back into the amp .... 'Scope the vas and other stages to see what's going on inside. You can then tell if you tend to overdrive the vas if you turn it up some into a dummy load. I'm sure you don't feel like listening to tones up high. That & you'd take out mids and tweeters doing that.
-Chris
There are numerous possibilities here, but your comment that you used a current mirror for the differential is interesting. I avoid using active loads whenever possible because they tend to produce an etched sound. Those who think measurements are the be-all, end-all love current mirrors and other, similar circuits. Those who actually listen to their systems tend to view them with a great deal of suspicion.
The usual reason people use an active load, whether current sources or their cousins, current mirrors, is to get as much gain as possible out of a stage. That's frequently because people intend to turn around and burn off that gain as negative feedback. High rates of negative feedback are also associated with poor sound quality.
Inevitably, someone will pop up at this point and start screaming about how it lowers distortion, etc. Your proper response to such a post is to ask them how often they listen to live, unamplified music. Amplified music doesn't count--it has to be unamplified, elseways you're listening to the PA system, and PAs make lousy stereos. Technical perfection that doesn't sound like real music isn't perfection, is it?
If you can't find evidence of actual misbehavior in the circuit, you might try using a resistive load and a lower rate of feedback. It might, possibly just cure the problem.
I'll back out now and let the measurements uber alles crowd fill up the next fifteen pages.
Grey
The usual reason people use an active load, whether current sources or their cousins, current mirrors, is to get as much gain as possible out of a stage. That's frequently because people intend to turn around and burn off that gain as negative feedback. High rates of negative feedback are also associated with poor sound quality.
Inevitably, someone will pop up at this point and start screaming about how it lowers distortion, etc. Your proper response to such a post is to ask them how often they listen to live, unamplified music. Amplified music doesn't count--it has to be unamplified, elseways you're listening to the PA system, and PAs make lousy stereos. Technical perfection that doesn't sound like real music isn't perfection, is it?
If you can't find evidence of actual misbehavior in the circuit, you might try using a resistive load and a lower rate of feedback. It might, possibly just cure the problem.
I'll back out now and let the measurements uber alles crowd fill up the next fifteen pages.
Grey
I scoped at the VAS collector and it looks just great but turning off the power to the output stage did make it spike a bit.
I decided to add some CC across the collector and base of the VAS.
It now sounds appropriately balanced.
Greg, I'm really glad you posted about current mirrors.
How about using cascode transistor loads like in the Leach?
Where do you find live music without PA's these days?
Even acoustic instruments are amplified at concert halls.
Tom
I decided to add some CC across the collector and base of the VAS.
It now sounds appropriately balanced.
Greg, I'm really glad you posted about current mirrors.
How about using cascode transistor loads like in the Leach?
Where do you find live music without PA's these days?
Even acoustic instruments are amplified at concert halls.
Tom
Hi, Grey,
This is a never ending story, between listening argument and measurement argument
Have you ever look to Electrocompaniet schematic, like AW120? It's quite complex, using mirrors, but its something. The RE degeneration are big everywhere, quite strange, why one use current mirrors but put big degeneration at the same time?
This is a never ending story, between listening argument and measurement argument
Have you ever look to Electrocompaniet schematic, like AW120? It's quite complex, using mirrors, but its something. The RE degeneration are big everywhere, quite strange, why one use current mirrors but put big degeneration at the same time?
tmblack said:
Hi Tom !
You ran into some typical trap. You increased openloopgain to
lower thd, but at the same time you reduced openloopbandwidth...
This results typically in exactly this sound you described.
You can remove this by adding a resistorload to the vas, try a 22k
to ground. Check your amp that it has openloopbandwidth flat
up to ~20khz.
Don't forget, low THD only means low distortion on sinus-signals.
Often low THD results in high distortions for complex signals
like music. But an amp is intended for music...
This crisp sound is normally heavy intermodulationdistortion,
it's not caused by instability.
Mike
No one seams to have popped up so I thought I would. NFB does lower distortion, etc.
Quite often down my local Jazz pub.
Just to apply some balance here:
If you really understand what you are doing...
a) Feedback is extremely good - as much as possible.
b) Bandwidth beyond the audible range is unecessary
c) Miller compensation is just fine, better than many methods.
I'm not trying to be controversial but rather I wouldn't want budding designers to adopt assumptions that limit the potential of their designs.
Hi Tom !
There are programs for this, but i don't know jack...
Maybe someone else can tell ?
I'm not sure if this kind of problem shows up in normal IM-tests,
i tried to reproduce this problem in sims, but with the normal
2freqs-im test it did not show up.
But you could try adding such a resistor in the vas, if it's this
OLBW-problem, the sound chould change immediately.
Mike
There are programs for this, but i don't know jack...
Maybe someone else can tell ?
I'm not sure if this kind of problem shows up in normal IM-tests,
i tried to reproduce this problem in sims, but with the normal
2freqs-im test it did not show up.
But you could try adding such a resistor in the vas, if it's this
OLBW-problem, the sound chould change immediately.
Mike
Re: Re: What do uncompensated amps behave?
Mike,
This is misleading. The open loop bandwidth is reduced, but only because the gain at frequencies below the original cross-over frequency is increased. The gain above the original cross-over frequency is unchanged. So overall, there is more gain available between DC and the original cross-over frequency.
Jan Didden
MikeB said:
Mike,
This is misleading. The open loop bandwidth is reduced, but only because the gain at frequencies below the original cross-over frequency is increased. The gain above the original cross-over frequency is unchanged. So overall, there is more gain available between DC and the original cross-over frequency.
Jan Didden
Hi Tom,
Not neccessarily. Current sources help isolate the signal from supply noise and make the circuit more stable if designed properly. Degeneration helps linearise the circuit and burn off extra gain so the amount of feedback can be reduced. Too much feedback absolutely sounds bad.
Besides, I really like all the lights from my LED biased current sources.
-Chris
Not neccessarily. Current sources help isolate the signal from supply noise and make the circuit more stable if designed properly. Degeneration helps linearise the circuit and burn off extra gain so the amount of feedback can be reduced. Too much feedback absolutely sounds bad.
Besides, I really like all the lights from my LED biased current sources.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.