Hi there,
I'm working on a NAP250 which has a bit of an instability problem (likes to oscillate at a low level around 3MHz). After much time trying to find the problem in the output stage it turns out it's actually the regulator producing the problem.
Working from Neil McBrides schematic: http://www.neilmcbride.co.uk/output-reg2.pdf
The main problem seems to be with the negative regulator on one channel although the noise finds it's way everywhere. I started getting a bit frustrated and poked around with a low value cap (say 1nF), bypassing resistors while watching the output on the scope. The result being that if I stick a cap across R110 (tail resistor for LTP) or R115 (bottom of feedback/voltage setting potential divider) the oscillation disappears.
What I don't know is what side effects adding bypass caps here will have, which is where all you clever guys come in!
Is there a better way to stop the oscillation?
Cheers for any help,
Chris
I'm working on a NAP250 which has a bit of an instability problem (likes to oscillate at a low level around 3MHz). After much time trying to find the problem in the output stage it turns out it's actually the regulator producing the problem.
Working from Neil McBrides schematic: http://www.neilmcbride.co.uk/output-reg2.pdf
The main problem seems to be with the negative regulator on one channel although the noise finds it's way everywhere. I started getting a bit frustrated and poked around with a low value cap (say 1nF), bypassing resistors while watching the output on the scope. The result being that if I stick a cap across R110 (tail resistor for LTP) or R115 (bottom of feedback/voltage setting potential divider) the oscillation disappears.
What I don't know is what side effects adding bypass caps here will have, which is where all you clever guys come in!
Is there a better way to stop the oscillation?
Cheers for any help,
Chris
Ooops, it was late and I wasn't looking at the circuit board properly.
It wasn't R110 and R115, it was R109 and R111 I was bypassing. (In fact I'm actually looking at the negative side so R209,R211 but the circuit is symmetrical).
Bypassing R211 increases the Miller compensation I assume and gives the desired effect by reducing the gain at HF. Cap across R209 does the same thing.
Neither of them are a good solution though.
So the question is, where is the best place to put some decoupling or compensation to get the gain at HF down.
It wasn't R110 and R115, it was R109 and R111 I was bypassing. (In fact I'm actually looking at the negative side so R209,R211 but the circuit is symmetrical).
Bypassing R211 increases the Miller compensation I assume and gives the desired effect by reducing the gain at HF. Cap across R209 does the same thing.
Neither of them are a good solution though.
So the question is, where is the best place to put some decoupling or compensation to get the gain at HF down.
What amplitude are we talking here. 'Scope across C205 directly, also connect scope probe tip and ground wire together and then connect both to zero volt line, never anywhere else particularly if 'scope earthed. Make sure you only connect the scope ground lead to the zero volt line when taking readings. Sounds mad but can you still see oscillation. Is the 3Mhz noise higher or lower in amplitude at the power amp end. Is the "front end" a linear or SMPSU.
Regards Karl
Edit, I hope it makes sense, it does not read very well.
Regards Karl
Edit, I hope it makes sense, it does not read very well.
Takign the +ve regulator as an example...
An easy thing to try is adding 1k in series with Q104 base, from the wiper of VR102. IOW a simple base-stopper.
Next try adding a cap, say 10uF/63V between the VR102 wiper and the output. This reduces noise-gain dramatically by reducing gain to 1x across the audio band, which helps stability and improves regulation.
Also take a good hard look at the parts choice for C105. This wants to be slightly lossy - Naim use electrolytics here, and when these get old, or if you substitute film types, you get exactly the oscillation you're seeing. You might want to experiment with 10uf with 0R5-2R7 in series with it.
An easy thing to try is adding 1k in series with Q104 base, from the wiper of VR102. IOW a simple base-stopper.
Next try adding a cap, say 10uF/63V between the VR102 wiper and the output. This reduces noise-gain dramatically by reducing gain to 1x across the audio band, which helps stability and improves regulation.
Also take a good hard look at the parts choice for C105. This wants to be slightly lossy - Naim use electrolytics here, and when these get old, or if you substitute film types, you get exactly the oscillation you're seeing. You might want to experiment with 10uf with 0R5-2R7 in series with it.
I've mainly been measuring across C205 and yes, I've only ever connected the 'scope ground to ground.
It's a linear power supply. At it's worst the oscillation is about 1Vp-p at the output of the regulator. You see the most noise at the output of the regulator. There are four regulator channels in the amp, all show some noise but I'm not sure whether it's just being generated in one channel that seems worse and propagating round or whether all of the channels have some tendency towards oscillation.
It's a linear power supply. At it's worst the oscillation is about 1Vp-p at the output of the regulator. You see the most noise at the output of the regulator. There are four regulator channels in the amp, all show some noise but I'm not sure whether it's just being generated in one channel that seems worse and propagating round or whether all of the channels have some tendency towards oscillation.
1volt pk/pk, thats a lot, is that directly on C205. When you say 4 regulators do you mean a pos and neg one for each channel. i.e. your circuit twice over. Is it clean on C105?, the pos rail. As Martin hints at, C205, have you soldered another cap across it to check, anywhere from 10 to 100mfd ok. I guess if you can eliminate it by doing as you said previously it has to be PSU related. It may be worth dropping the value of R211 and I see R111 is different, not the same value. To me that points to a "twitchey" bit of design, it should not be that critical compensating a regulator. 3mhz is pretty high, what happens if you touch R213 with your finger with it on, does it alter anything ? Be careful and only use one hand, other out of way 
From what I've seen so far Naim circuits generally seem quite 'twitchy'.
Cap from base of Q104 to output made it worse, doubled amplitude of oscillation. Putting a base stopper in would be annoying due to layout, would have to cut the track, mount on the bottom etc.
I've just managed to kill the regulator somehow with my experiments. Need to work out what's happened before I can get back to finding the oscillation.
Cap from base of Q104 to output made it worse, doubled amplitude of oscillation. Putting a base stopper in would be annoying due to layout, would have to cut the track, mount on the bottom etc.
I've just managed to kill the regulator somehow with my experiments. Need to work out what's happened before I can get back to finding the oscillation.
Ok, some interesting findings.
Me 'killing' the regulator was actually it shutting down, I'm assuming due to the current limit. I think it was just inrush current when connecting an extra cap across C205 that tripped it. Interestingly it shuts down the supply until you power cycle the amp (and let reservoir cabs discharge).
Anyway, connecting an extra 10uF cap and powering up seems to get rid of the oscillation so I can only assume that the currently fitted 10uF 63V electrolytic is dry and dead. I'm going to re-cap it with some 47uF 63V caps I've got lying around and see how that goes.
Guess I should consider doing the input caps too (C101/201)
Thanks for your help so far guys.
Me 'killing' the regulator was actually it shutting down, I'm assuming due to the current limit. I think it was just inrush current when connecting an extra cap across C205 that tripped it. Interestingly it shuts down the supply until you power cycle the amp (and let reservoir cabs discharge).
Anyway, connecting an extra 10uF cap and powering up seems to get rid of the oscillation so I can only assume that the currently fitted 10uF 63V electrolytic is dry and dead. I'm going to re-cap it with some 47uF 63V caps I've got lying around and see how that goes.
Guess I should consider doing the input caps too (C101/201)
Thanks for your help so far guys.
Sorry, I'd misread that you are working on an existing NAP250, I saw'mcbride' and thought 'clone'.
Anyway - with 250s and 135s replacing that output 10uF cap is the very first thing to swap when they oscillate. These caps die with age, and when the ESR gets too high - oscillation time.
Glad you've got it workng!
Anyway - with 250s and 135s replacing that output 10uF cap is the very first thing to swap when they oscillate. These caps die with age, and when the ESR gets too high - oscillation time.
Glad you've got it workng!
It's a bit of a wierd one, it's what you could call a 'grey' NAP250, not going to go into too much detail on that. It's a good 10-15 years old though.
Originally came to me having taken out the Zobel on one channel (properly cooked). Fixed that but still had a bias issue, drivers were running very warm. That turned out to be one of the diodes in the current source for the LTP then I've been hunting this oscillation for a while.
It's going into a mates studio to run a pair of Tannoy Gold Spots.
First I need to build some accessories for it. Soft start, loudspeaker muting for switch on/off. Possibly a balanced line receiver.
I'll let you know the final results.
Originally came to me having taken out the Zobel on one channel (properly cooked). Fixed that but still had a bias issue, drivers were running very warm. That turned out to be one of the diodes in the current source for the LTP then I've been hunting this oscillation for a while.
It's going into a mates studio to run a pair of Tannoy Gold Spots.
First I need to build some accessories for it. Soft start, loudspeaker muting for switch on/off. Possibly a balanced line receiver.
I'll let you know the final results.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.