LM3875 Gainclone-current state of the art survey

[Dr.H]

All,


Having built my first channel of the gainclone (and been very surprised/happy with its performance), I would like to establish what the current "best practises" are. In particular, what is the thinking on the following:

1. Use of bridge diodes vs seperate diodes-Are the fast soft recovery ones recommended and what specifically? What about snubbers? RC or just C? Suggested values?

2. Is the bypass of the PS caps (1000uF per pin (+ and -), Panasonic FC) recommended, if so with what value, type

3. Is inverted still regarded as the better option? What about the low input impedance?

4. Pin 7 to ground still recommended?

5. What value feedback Resistor and type (metal film, carbon) is being used. I use 220k metal film

6. What about the input resistor? I use 10k metal film. Views on volume pot-Linear vs log and what sort of values suggested-100k, 470k, 50k?

7. What about the use of input buffers-JFET vs tube based

7b. What about biasing into class A with a 1.5k resistor to V-? Anthony Cockrill mentions a CCS (constant current source)somewhere. Anyone tried this?

8. Has anyone experimented with speaker cable on the beast

9. Finally, what is the latest view on chips-LM3875vs LM1875 vs TDAxxxx etc

Thanks for any input. Seems we should be collating this information somewhere. My gainclone setup summarised below:

300VA, 25VAC (34 DC or so), 32mV DC offset, VERY cheap 450k linear pot, 12cmx7cm aluminium heatsink, 1000uF Panasonic FC caps, 25A bridge diodes, LM3875, inverting configuration, pin7 to ground, 220k in feedback loop, 10k in to pin 8. No input cap, fed straight from my CD player.


Regards
Ryan

[Peter Daniel]

1. Definitely separate diodes are recommended, as they affect sonics in a big way. I tried few different types and like MUR860 the most. Using separate bridges for ea. rail is also recommended, as one bridge doesn't sound good. Didn't try snubbers, but I'm using 4.7u BG N cap right after bridges in a sep PS. When using separate PS, the umbilical cord affects the sound as well and it pays to use a good one. Mine is made out of Cardas hook up wire (19ga., 8 wires).

2.Bypassing main PS caps maybe good or bad, it's up to you how you like it. This is one area when experimenting is recommended. Personally, I don't care much for bypassing; it seems to make the sound artificial too much.

4. Pin 7 directly to ground sounded better for me than using resistor in that place.

5. I decided on 250K feedback resistor as this gives me slightly more gain, yet sounds better than 300k. Also feedback resistor type is very important to good sound and I chose Caddock MK132 and prefer it to Riken. Caddock makes the sound more liquid and 3-dimentional comparing to Riken, which tends to be a bit dry (in my setup). The proper resistor choice will greatly depend on your setup.

6. I also use 10k Riken for input resistor and 50k Nobel attenuator. Tied 100k but 50k works better. Also it is a log pot, but the control is very good, never tried the linear pot.

7. I tried biasing the output with 1k5 resistor from the rail, but disconnected it quickly as the sound became worse.

8.Didn't experiment much with speaker cables, but also didn't find problem with anything I tried. Lately, I've been using copper foil for the cables (out of Alpha Core inductors) and I like it a lot; transparent, clear and detailed.

9. Tried TDA7294, was tweaking it for few days, couldn't make it sound good. Put LM3875 in a same chassis, sounds much better.


As a side note, I can add that the chassis construction will also affect the final sound of your GC. This is actually the are were most experimenting is recommended.

[Nuuk]

I hate to say this Dr H but all the answers that you seek are contained in the the thousands of posts here in the Chip Amps threads.

There a lots of differing opinions on circuits, components and even methods of building. The only real way to know who or what is right, is to experiment for yourself.

[Dr.H]

Peter: Thanks for the response, very useful and food for thought. Given your experience in this area, your post is especially valuable. I assume that you tried the simple bridge diodes. My chassis will be a relaively cheap steel casing. Do you think it will benefit from damoming or is there there anything else I can do for improvements?

Nuuk: Yes, I saw that some threads run to the 100's! I wish the search engine would poick up specific portions of a thread rather than just finding the overall thread, which one then has to search from post 1 to n to find the specific ietm you're looking for...

Thanks again.
Ryan

[pinkmouse]

Originally posted by Dr.H
I wish the search engine would poick up specific portions of a thread rather than just finding the overall thread, which one then has to search from post 1 to n to find the specific ietm you're looking for...

Try the "Show results as Posts" button on the search page.

[Dr.H]

Thanks, great advice.