"Yuanjing" Gainclone 3886 - eBay amazing value ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
HI all
got the opa2134 installed on one channel I know this is very well regarded . I couldn't tell much difference but then i havnt done much critical listening for a while .perhaps more detail such as fingertaps on the body of some acoustic guitar had more reverb/air to the decay of the sound.May leve the ne5532 on as this will be woofer amp in a three way.I think the input cap change to 4.7uf makes a bigger difference.Dont seem to be able to get carbon comp resistor here in Oz.might have to ebay from usa.
 
got the opa2134 installed on one channel I know this is very well regarded . I couldn't tell much difference but then i havnt done much critical listening for a while .perhaps more detail such as fingertaps on the body of some acoustic guitar had more reverb/air to the decay of the sound.

Try the LM3886-side lifted-ground mod (the last one) - a clean ground makes a huge difference to detail and transparency, especially with premium op-amps. Without a clean ground, the NE5532 sounds as good or better than almost every other opamp. The only downside to the said mod is that the output offset has jumped from 0.3/0.5 mV to something like 90/130 mV, which I can live with.


May leve the ne5532 on as this will be woofer amp in a three way.I think the input cap change to 4.7uf makes a bigger difference.Dont seem to be able to get carbon comp resistor here in Oz.might have to ebay from usa.

For a woofer amp, no further mods (except the input ground lift and larger input cap) are necessary. The carbon-composition is optional - you may get a similar improvement in sonics by using just about any low-inductance resistor, like a metal-oxide or similar. Perhaps the higher value (27k or 33k) actually plays the important role. Carbon-comps also have a voltage-dependent resistance effect, but it probably isn't important at low swings.
 
The fully-modded gainclone with better opamps sounds sublime in the mids and highs - you can use maybe just one for the mid-range and tweeter combined, with a simple capacitive crossover for the tweeter. The Mauro Rev C is an even better candidate for this, but the current group buy may be expensive.
 
Wow, I can't believe I'm not the only one getting entertainment out of this cheap amp board!!

Linux, can you recommend the best/cleanest way to bypass the NE5532 in the YaunJing board? I know I don't need the extra gain (is actually detrimental in my case). If you were looking to get as clean of sound out of this cheap board without really putting any $$$ into it, which mods would you make? I'm not really worried about bass output, as I'll be running a sub with a separate amp.

Great stuff guys!!!
 
Linux, can you recommend the best/cleanest way to bypass the NE5532 in the YaunJing board? I know I don't need the extra gain (is actually detrimental in my case). If you were looking to get as clean of sound out of this cheap board without really putting any $$$ into it, which mods would you make?

If you're bypassing the op-amp, then it will probably sound OK, but not great, without any mods.

However, it would help to isolate the signal ground for the LM3886 input shunt resistor and feedback network from the ground plane around it (which carries large return currents from the speaker), and jumper it to a ground section which doesn't experience significant ground-bounce. A good candidate for that is the section of the ground adjacent to the op-amp.

(Edit: mods for this to be detailed later.)

Optional:

1) Upgrade the Wimas to (preferably larger valued) MKP or similar input-coupling caps.
2) Try carbon-composition resistors for the input shunt resistor as well as feedback resistor of the LM3886. I got good results with 27k, 1/4 W Allen-Bradleys at both locations.
 
Last edited:
If you're bypassing the op-amp, then it will probably sound OK, but not great, without any mods.

However, it would help to isolate the signal ground for the LM3886 input shunt resistor and feedback network from the ground plane around it (which carries large return currents from the speaker), and jumper it to a ground section which doesn't experience significant ground-bounce. A good candidate for that is the section of the ground adjacent to the op-amp.

(Edit: mods for this to be detailed later.)

Optional:

1) Upgrade the Wimas to (preferably larger valued) MKP or similar input-coupling caps.
2) Try carbon-composition resistors for the input shunt resistor as well as feedback resistor of the LM3886. I got good results with 27k, 1/4 W Allen-Bradleys at both locations.

I look forward to your detailed mods. Any specific recommendations on bypassing the op-amp? I know that may seem obvious, and there is of course several places that look obvious to me, but I'm assuming there are probably good ways to do it and better ways.

Thanks MUCH!!!
 
Linux, I bypassed the buffer by just wiring into the existing wifa and removing the NE5532's and it's still not the quietest amp. I'm looking forward to any details you come up with on isolating the input ground. I know you detailed the proceedure for use with the similar board along with the input buffer mod but I'd love to see specifics without the buffer.

Any recommendations for a source for replacement input caps, as well as supply decoupling caps to solder directly to the amp power pins? Seems everyone has their favorit caps for those uses, and a google provides so many options it's dizzying.
 
linuxguru reckons performance is best with all the mods and op amp intact. im going to try carbon resistor mod and second lifted ground mod.ill then try opa2164 opamps again.

Sorry - preoccupied with other matters. Actually, I'm not sure if it's better with or without the opamp in the signal path - I'd guess that the latter is better. However, if you do want to keep the op-amp, then both lifted ground mods are recommended if you want to use a fast, high-performance op-amp like the LM4562 or LT1208. Without the second lifted-ground mod, the ground-return currents from the speaker will contaminate the ground reference for the signal/feedback networks near the LM3886. The downside of the second lifted-ground mod is that the output offset increases from mVs to ~100 mV.

Again, regarding the carbon resistor - it might well be the increased value (27k vs. 22k), rather than the type of resistor, that is relevant to improving the sonics. It's probably worth experimenting with different values (up to say 47k) and types (Dale brown MF, Allen-Bradley CC, and maybe boutique types like Shinkoh, Caddock, etc.) at these locations. I used Allen-Bradley 27k because I just happened to have them lying around.
 
my second unit has arrived.wima caps have differnet layout of printing but still say 1.8uf (not listed on wima website!)it has the non isloated chips on this time complete with screw, tape and plastic bush.
do i still use thermal paste?

Regarding the Wimas, they could be counterfeit. It's not difficult to remove and upgrade them with known good ones. The LM3886TF (isolated case) are the preferred ones, because it simplifies heatsink mounting - you can use a single heatsink and also bolt it to the chassis easily. With the metal-tab LM3886, the paste is optional if it has the soft gray or pink rubbery isolating tape.
 
Linuxgurru sorry if misquoting re opamp bypass improvement.
I may well try bypass in future when I have ESP electronic crossover as they have an output buffer.IT may be unnecessary to have the output buffer running straight into another buffer.
what do you think?
 
Last edited:
Got my second board with full mods up and running sounds great! but when i touch the metal of of the volume pot (alps blue) i get mains hum?!
Hey linuxguru i notice on the second mod photo for the jumper lead it also connects to the gnd pin of the input connecter but this is not mentioned in the text is this something else to be done?
 
when i touch the metal of of the volume pot (alps blue) i get mains hum?!

It's some kind of introduced ground-loop, with your body being part of the pick-up loop. You may have to isolate the power ground from the chassis ground to get rid of it.

Hey linuxguru i notice on the second mod photo for the jumper lead it also connects to the gnd pin of the input connecter but this is not mentioned in the text is this something else to be done?

Good catch! Yes, on one of my boards I isolated the ground pad of the input connector by using an exacto knife on the copper on both sides (component and solder side). I then bridged the isolated ground pad to the lifted-ground. This mod is purely optional, but makes it more convenient to connect and disconnect the input shield ground without soldering/desoldering the shield wire at the upper lead of the 10 ohm ground-loop breaking resistor.
 
Last edited:
I bypassed the op amp by connecting the input to the second Wima, and I removed the op amp alltogether. I left the other channel stock. There was a little improvement in the noise floor, but not enough to make it usable (in my opinion). Running the output from my mp3 player instead of the input forom my preamp improved the noise floor a lot but didn't clear it up completely (suggesting the noise floor is a grounding issue????).

When I did the second ground mod and used 27k resistors, the noise floor got worse, and subjectively the sound quality got worse as well. Of course, without being an expert at this stuff, I don't know if I ran it all the best possible way. Just desoldered the two resistors, put the two new ones in and scraped off the tracings to the required grounds and soldered jumpers to a spot near the op amp.

I'd really love to get this thing to a usable sound quality level as a couple of them would be great for a project I'm working on that is cost sensitive.
 
When I did the second ground mod and used 27k resistors, the noise floor got worse, and subjectively the sound quality got worse as well. Of course, without being an expert at this stuff, I don't know if I ran it all the best possible way. Just desoldered the two resistors, put the two new ones in and scraped off the tracings to the required grounds and soldered jumpers to a spot near the op amp.

OK - I guess you removed the DC blocking cap and bridged the resistor to the adjacent resistor as well? You need to isolate both the points where the resistive network near the LM3886 is grounded. For me, it improved the sonics audibly, but I've retained the op-amp pre/buffer.

The mods can only go so far - the original ground-plane based design is nonsense. There's no way to avoid ground-bounce in the original design. There's now an improved version of this board on EBay, with blue solder-mask and an LM833 buffer. It has better grounding, from the EBay pictures. However, it's priced at $45 assembled, which makes it less interesting - for that price, a discrete design like the Symasym/Goldmund is preferable.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.