NAP250 Reg Board O/P Capacitor Question

AKprentice · #1 21-01-09, 01:42 PM

Has any one tried a plastic film cap on the output of the NAP250 reg board - if so was any additional series R or L used to maintain reg stability.

It seems that adding a 1 ohm series R (as recommended for preamp decoupling) will not be effective for a power amp, because of the much higher currents involved. I was wondering whether a small series inductor might do to the trick.

Any thoughts are welcome

Andy

zener · #2 21-01-09, 01:50 PM

I found removing the reg board and putting something else in its place greatly improved the amp , it was part of my DIY amplifier journey.

AKprentice · #3 21-01-09, 02:42 PM

Zener,

I have seen many comments on the improvement from removing the reg board, but from what I understand, this requires investment in a new transformer to avoid over-voltaging the power-amp boards. Have I got this right?

Investment in a new transformer is not an option right now.

Andy

zener · #4 21-01-09, 03:00 PM

I'l get back to you , with more info...

martin clark · #5 21-01-09, 03:02 PM

There are other things you can do to the reg boards that should boost performance nicely..

More soon.

AKprentice · #6 21-01-09, 03:16 PM

Martin,

I have been thinking about improvements to the reg board myself. This seems an are that has not been exploited very much so far.

My initial thoughts were adding current sources for the LTP and the Zener. I look forward to hearing your ideas.

Andy

Pete MB&D · #7 22-01-09, 12:09 AM

Hi,

I have a couple of 1.8uF film caps on the reg outputs on my clones it does help I seem to remember, no series resistior and its own return to 0V.

Pete

zener · #8 22-01-09, 01:31 AM

I did what I think martin is going to suggest , removed all the resistors and replaced them with 1 watt rated of the same ohmic value and fitted 250 volt rated Electrolytics.

PigletsDad · #9 22-01-09, 01:51 AM

Current source for the LTP won't make much difference - it operates at fixed voltage, so CMRR isn't an issue, so I doubt the improvement will be worth the trouble.

Current source for the Zener is a very good idea.

You could also put a little RC low pass network after the Zener, to further reduce HF noise, both from the raw supply and internally generated by the Zener.

Resistors in series with the output caps don't see any DC current, so 1 Ohm would be OK, if you want to play round with film caps. Actually I would probably go for a lower value, say 0.22Ohm. An inductor without damping resistor in series with a cap is asking for trouble - that is just a resonator circuit!

martin clark · #10 22-01-09, 02:18 AM

What I was going to suggest, but haven't had a chance to check for stability issues, is a feedback cap in parallel with the 'upper' feedback resistor (R114, 5K6ohms on Neil McBride's schematic here ).

I would try a cap in the range 1uF-4u7; preferably at the larger end. It needs at least 50v rating. This would reduce noise gain and should improve the regulation and dynamic performance somewhat.

You could actually take this to the wiper of the voltage set pot, but I wouldn't: just adding the cap across R114 leaves a bit of resistance in series with it which should help ensure stability.

I haven't had the chance to test this, and would caution that it really would need watching with a source, a scope running the amp into a dummy load for a while to ensure it behaves properly! Adding the cap requires that the whole error amp / reg output loop is unity-gain stable, and that's not guaranteed at all.

PigletsDad · #11 22-01-09, 03:54 AM

That sounds an excellent idea, but I suspect will require some fiddling with the loop stability, as the noise gain goes from 4 to 1. So I expect that C104 and/or R111 would need tweaking. My first guess would be to increase C104 to about 2nF, and reduce R111 to 160Ohms; this would reduce the first pole by about a factor 4, and leave the zero roughly where it was.

For testing, put square waves through the amp, into a resistive dummy load, and look for ringing and overshoots on the voltage rails.

LesW · #12 22-01-09, 04:17 AM

Here's one I made earlier: http://img510.imageshack.us/img510/3...dnapreglv5.jpg

Would some kind, erudite soul care to run this on a sim package.? Mine has the hiccups.

Thanks, Les

PigletsDad · #13 22-01-09, 04:36 AM

Les,

I'd love to do this, but won't be able to get to it till next week at the soonest; I have to get a couple of things out the door (quote and a report) today.

My eyeball guess would be that you need to slug the dominant pole down, probably by increasing the 47pf, and maybe put a resistor in series with it to get pole/zero compensation.

LesW · #14 22-01-09, 04:51 AM

Thanks PD, I'd value your input/comments. I added the 47pF as a token pole compensation as I happened to have some silver mica caps to hand but fully realised it wasn't the be-all and end-all to slugging the output triple. I did yesterday, mock this up out of a redundant NAP reg PCB and the DC conditions are pleasingly stable - the voltage settling to within 0.1V of the set value within seconds of being energised. There may be chance to connect this to an NCC200 board this weekend and run it it up on the 3582A.

ced1 · #15 22-01-09, 05:26 AM

Id say the obvious things are to:
a) stick a decent current source on the long tailed pair. Someone did this years ago and reported an improvement.
2) stick in a high voltage gyrator to feed the reference and diff input only.
3) Replace the reference zener with an lm329dz and change the feedback resistor values accordingly (its a 6.9v ref not a 10V). Probably stick with a resistor in front of this as you're now running off a regulated rail. Bit of filtering of the reference voltage also. Say 10uf polyester + 500ohms. Tweak the impedances to match both sides of the diff as we are bipolar after all.
To be honest, I reckon regulating the front end will give the biggest improvement by a fair margin judging from experience.

If someone wants to try it, I'll post a schematic I have somewhere with some values.

cheers
Ced