advertisement


Cheap Balanced Power Amplifier - Chip or Class D

JemHayward

pfm Member
I'm currently running my Martin Logan Summits from my much loved EAR 890 Valve amp. I'm hoping (one day...) to take delivery of a set of John Westlake's VFET power amps which are going to be developed specifically for this type of speaker, so I have a final solution, but I would quite like to release the invested funds in the EAR for other things, so I was thinking about building a small power amp to use in the interim period.

My kitchen system is a three way active speaker and I use two T-amps for the mid and treble, and a pair of Gainlones that I built maybe ten years ago, that used to drive my Quad ESL57s in my TV system.

I've tried the Gainclones on the Summits and they sound good, but I'm not sure how well t-amps will cope with the extremely low impedance at high frequencies.

I'd like to build an amp that has a balanced input and is relatively simple to construct. Ideally I'd like to re-use some PSU components I have in stock as well, to keep costs down.

I'm sure someone is going to suggest N-Core but that is too much money.

So, should I just build a balanced Gainclone, or should I be investigating class D - T-amp type of technology, as they seem to be available as ready built boards quite cheaply.
 
Take a step back from N-core to UCD. MORE than good enough.

Hypex seem to be selling off the UCD32MP which appears to be a combined SMPS and two channel amplifier...

Is this any good? Has anyone any experience? It would seem to be a simple and cheap way of getting a half decent power amp, assuming it is half decent!
 
Hypex seem to be selling off the UCD32MP which appears to be a combined SMPS and two channel amplifier...

Is this any good? Has anyone any experience? It would seem to be a simple and cheap way of getting a half decent power amp, assuming it is half decent!

I decided it was so cheap, I'd just give it a go... so I'll report on the result soon.
 
The UcD module has arrived. It has the required low impedance balanced input I need, but is DC coupled. My "preamp" is DC coupled too - theoretically no digital music file should have DC on it, but I've been told that some poorly designed soundcards can end up outputting a 1 instead of a 0 when idle, thus putting DC on the output.

I'm tempted to take the risk as I don't want to add any more circuitry to the power amp on both cost and minimalism grounds, but I'm wondering about using input capacitors just in case.

If I go down this route, and want to block, say everything below 5Hz, will I need something implausibly huge?
 
I may also want to reduce the effective gain by attenuating the input as well, which would effectively reduce the input impedance yet further. How do I do the maths on this to get the right value? I recall it has Pi in it somewhere?
 
f= 1/ (2*pi*R*C) so C= 1/ (2*pi*R*f)

22uf would be a standard value that looks about right based on Julf's comment./ assumption of input impedance.
 
I may also want to reduce the effective gain by attenuating the input as well, which would effectively reduce the input impedance yet further. How do I do the maths on this to get the right value? I recall it has Pi in it somewhere?

Would you use either a series resistor or a voltage divider to attenuate the input? In that case it makes sense to put the capacitor before the attenuator, so that the capacitor "sees" a larger impedance (and can thus be smaller).
 
My preference would be to use a series fixed resistor of good quality, and then maybe a variable to 0v to allow me to trim the gain (as I'm using a MDAC and would want to keep the digital attenuation less then -20dB).
So, input board looks like 22uf capacitor in series with, say, 4k7 and then a 47k trimmer pot on each leg between input and 0v as I'm using balanced input. I'd then need to trim the pots together, which could prove interesting, but its a one off exercise, the put the lid on and forget.
 
Any thoughts on capacitor and resistor types (without going bonkers as this is a cheap amp project, not an ultimate amp project)?
 
My preference would be to use a series fixed resistor of good quality, and then maybe a variable to 0v to allow me to trim the gain (as I'm using a MDAC and would want to keep the digital attenuation less then -20dB).
So, input board looks like 22uf capacitor in series with, say, 4k7 and then a 47k trimmer pot on each leg between input and 0v as I'm using balanced input. I'd then need to trim the pots together, which could prove interesting, but its a one off exercise, the put the lid on and forget.

Personally one trimmed I would measure the setting and swap in good quality resistors.. So simply using the trimmers to work out the required value
 
My preference would be to use a series fixed resistor of good quality, and then maybe a variable to 0v to allow me to trim the gain (as I'm using a MDAC and would want to keep the digital attenuation less then -20dB).
So, input board looks like 22uf capacitor in series with, say, 4k7 and then a 47k trimmer pot on each leg between input and 0v as I'm using balanced input. I'd then need to trim the pots together, which could prove interesting, but its a one off exercise, the put the lid on and forget.

In that case the 22uF cap is unnecessarily large - you can get away with a 6.8 uF one...
 
say, 4k7 and then a 47k trimmer pot on each leg between input and 0v as I'm using balanced input. I'd then need to trim the pots together, which could prove interesting

With a balanced input you only need (or want) the one trimmer between the inputs, after and directly between the two 4K7s - not from each to the screen connection bond. So two trimmers total for 2 channels.
 
Having spent my lunch hour reading what I could fine on the internet, I think that the UcD 30 modules that make up my amplifier are relatively low gain already, as they don't have the input buffers that the UcD180 etc, have, so I'm hoping they may not need any input trimming at all, so I think I may start off with just DC blocking caps. 22uF would seem to be a safe bet. I'm not sure if I have anything useful "in stock" so would these be ok?

http://uk.rs-online.com/web/p/produ...|prd|6844884&gclid=CKC_8p3Vr88CFWIq0wodhkgDUg

Not cheap, as I need 4 of them, but not silly either.
 
There are people who obsess about capacitor types, and prefer polystyrene (not available in the size class you are looking at) or polypropylene.
 
With a balanced input you only need (or want) the one trimmer between the inputs, after and directly between the two 4K7s - not from each to the screen connection bond. So two trimmers total for 2 channels.

Ah, that makes sense... I'm still thinking single ended... but now thinking I may not have excessive gain anyway, as it seems that the gain in the UcD topology comes mainly from the input buffer, that I want to avoid.
 
Update:
Hypex unit arrived safely, another fine case from Douk and I now have a power amplifier!
Unexpected costs pushed the budget up, as I decided to invest in a 25mm hole punch to mount my XLRs and I decided that maybe a quartet of Vishay 22uf Foil caps to block DC on the input, even though the four caps are physically as large as the PSU/amp and cost (almost) half as much! I added in a couple of switches to make testing easier (not really necessary) and made up a perspex cover so I couldn't interface my fingers with 240v AC (I have a heart condition already!).
So, I now have a 30W (4 ohm) power amp with SMPS in a case featuring heatsinks (think lego version of a Krell) - even though no heatsinking is needed. However they do work with my Martin Logan Summits - which is quite an achievement! I've not yet tried hooligan levels (my day off is Monday!) but it copes better with the punishing HF impedance than my my NAP250 (it just shut down for a few minutes) or my Gainclones, which do start switching on and off when fed anything above 14kHz.
Wife is ambivalent - doesn't want to part with the valve amp, but likes the more "etched" sound...

If I can extract it from the main system I may takes some photos, but visually nothing special...
 


advertisement


Back
Top