Repurposing a UcD32MP SMPS.

Shoog

Member
2002-08-15 10:16 pm
Eire
Hi all,
I have a nice UcD32MP amp on my main system and I am very happy with its performance, good enough to replace all of my valve amps. However just before xmas I was hunting a small residual hum and I accidentally dressed the leads to close to the amp board and one of the opamps fried. I got a replacement and all was good again.
However I now have a top class SMPS left on the dead UcD32MP and am reluctant to just bin it. I was thinking I could re purpose it to drive a Gainclone.
I have one of these driving my woofers
lm3886 kit - Google Search
I am more than happy with the build quality of this kit and it has proved reliable and excellent quality. I also have some spare OPA637 chips to upgrade the buffer stage.
So my questions are:
- I am assuming that the Hypex SMPS has split output rails, is this correct ?
- How easy is it likely to be to fool the SMPS into powering up, is it just a matter of shorting the SMPS enable line ?
-the idea was to simply bypass the rectifiers on the kit and hook straight up to the big caps. However this will present a dead short to the SMPS which would need some sort of soft start circuit to overcome. Would it be more beneficial to simply leave out the caps altogether.

Are there other digital alternatives to the LM3886 which could marry with the UcD32MP SMPS.

All help appreciated.

Shoog
 

00940

Member
2003-09-03 12:17 pm
Liège
- The amp rails are +/-18VDC.
- You'll need to use the enable and maybe provide the correct pullup/down for the fault line coming from the amp. The pinout of all the connectors is given in the datasheet available on hypex website.
- I would avoid adding more big caps after the smps.
 

Shoog

Member
2002-08-15 10:16 pm
Eire
Thanks for the advice, it helped along the way.

Stripped out the SMPS and all that is needed to enable it is the SMPS enable shorted to standby 5V. Tested and seeing +/-18.9V on output so I have to assume that the SMPS is OK.

Ordered LM3886 kit. I have high hopes for this combination as a OPA627 buffered LM3886 with SMPS was about the best sounding gainclone that people achieved. So it will be interesting to see if it matches or betters the Hypex Class D amps.

Be a few more weeks before I get results.

Shoog
 

Shoog

Member
2002-08-15 10:16 pm
Eire
So I got the amp and after a few minor hickups got it built.
The SMPS works fantastically with it, helping to improve the bottom end no end. Its absolutely silent with nothing attached. The only issue I have is that the gain is significantly higher than any of my other amps which makes my Computer a very noisy source and has revealed a significant residual hum in my diy preamp. I will probably do a voltage divider at the front to tame it.
Comparing it to the Uxd32p and its good but not quite as good. Its a bit more shrill and not quite so natural sounding and the sound stage is a bit smaller. However its revealing things in the mix which were previously not obvious so detail is high.

Overall a useful exercise and I have a reserve amp now and proof that these SMPS can drive a analogue amp.

Shoog
 
Hi,
fine all went so well. If you trust high end, high price audiophile tests, the ultimate LM3886 was a parallel / bridged 6 chips per channel amp, driven by a SMPS, better than the same with a linear PS.
So your amp should be able to sound very well. Often different sound is a matter of taste, more than better or worse.
Could you post some pictures of your build and a link which amp kit you build? There sure is some interest in your construction!

Thank you and good listening
 

Shoog

Member
2002-08-15 10:16 pm
Eire
Added a bit of inductance paralled to a 10R resistor on the output and it seems to have calmed down the stridency a bit. Will have to get the components for a zobel as well. The kit is bare wire to the output and my speakers are all 4R so a heavy load for these types of amps.
Did a 220hz square wave test and its almost perfect with just a tiny bdelay on the rise and a slight overshoot on the decay. Also did a 20hz to 20khz sweep and this showed it was flat from 20hz up to 10khz and then it tailed off a bit, but no nasty resonances at all.

Here is a photo of the build, into a scappy old case. The kit is the standard LM3886 kit that is available from China and HK, very well executed and good quality parts for the price of €35.00. I replaced the 10,000uf input caps with DC chokes but I would image there is no noise coming from the SMPS that needs attention.

Shoog
 

Attachments

  • IMG_20200212_152804.jpg
    IMG_20200212_152804.jpg
    367.2 KB · Views: 80

Shoog

Member
2002-08-15 10:16 pm
Eire
Final comment on this. If anyone buys one of these lm3886 kits there are jumpers on it which are not well documented anywhere. There are two jumper cs and os. If cs is linked then the 4u7 cap in the signal path is bypassed giving a straight through signal from the input to the lm3886.
If the os jumper is linked then the input opamp buffer is included in the circuit. This fills out the sound considerably.
So the jumpers are not an either or situation, best results are achieved with both linked.

Shoog