Hi Linuxguru, Yes the little caps (with resistor like bands) are the 100pf suppression caps. Those Panasonic PSU caps are big value. I have used a couple of Elna 6800uF. Do you think I should fit bigger ones? I had it running without a case and it sounds very good although I only run it for a few seconds while I adjusted the dc offset. What sort of pre-amp would you recommend?
Hi Linuxguru, Yes the little caps (with resistor like bands) are the 100pf suppression caps. Those Panasonic PSU caps are big value. I have used a couple of Elna 6800uF. Do you think I should fit bigger ones? ... What sort of pre-amp would you recommend?
Thanks - 4700 to 10,000/50V should be sufficient. I chose the 22,000/50 because it was almost the same size and price as the 12,000/63. Any op-amp based preamp based on an OPA2134 or LM4562 should be good enough - there seem to be several such PCBs/kits on EBay.
I'm already using the LME49860 (higher-rail version of LME49720/LM4562) and it sounds great. I noticed that the improved version of the Yuanjing (blue PCB) on EBay uses the LM833 - it's certainly worth a roll, though I doubt that it can trump the LM4562.
I just completed the Symasym/Goldmund PCB and powered it up. I've just started listening to it, but here are my first impressions: sonically, it's a dead-heat between the modded Yuanjing with LM4562 buffers, and the Goldmund/Symasym. There are subtle differences, but it's too close to call.
Amp A - Yuanjing with LM4562: +/- 28V rails, 4700 uF Elna CE on each rail, 2.2uF and 3.3uF X2 MKP coupling caps, LME49860 preamp/buffer, all the mods previously described in this thread.
Amp B - Symasym/Goldmund: +/- 28V rails, 22,000 uF Panasonic TS-UP on each rail, 10 uF Nichicon Muse BP input coupling cap, no preamp, 470uF Rubycon YXA for supply filtering for the LTP and VAS sections (instead of 100 uF Panasonic FM supplied in the kit), 2sc5200/2sa1943 O/Ps, 2sc5171/2sa1930 drivers, 100 mA output-stage Iq per channel (Clas-AB).
Audible Impressions: Distortion and sonics are nearly identical. The Symasym has slightly tighter and better defined bass, while the Yuanjing has better detail in the upper-mids/highs. The Symasym is slightly darker and less grainy.
Conclusion: With a BoM of around 1/3rd the Symasym/Goldmund, the modded Yuanjing smokes the Symasym/Goldmund on value, and holds its own on all audible parameters, trailing only slightly on bass authority.
Amp A - Yuanjing with LM4562: +/- 28V rails, 4700 uF Elna CE on each rail, 2.2uF and 3.3uF X2 MKP coupling caps, LME49860 preamp/buffer, all the mods previously described in this thread.
Amp B - Symasym/Goldmund: +/- 28V rails, 22,000 uF Panasonic TS-UP on each rail, 10 uF Nichicon Muse BP input coupling cap, no preamp, 470uF Rubycon YXA for supply filtering for the LTP and VAS sections (instead of 100 uF Panasonic FM supplied in the kit), 2sc5200/2sa1943 O/Ps, 2sc5171/2sa1930 drivers, 100 mA output-stage Iq per channel (Clas-AB).
Audible Impressions: Distortion and sonics are nearly identical. The Symasym has slightly tighter and better defined bass, while the Yuanjing has better detail in the upper-mids/highs. The Symasym is slightly darker and less grainy.
Conclusion: With a BoM of around 1/3rd the Symasym/Goldmund, the modded Yuanjing smokes the Symasym/Goldmund on value, and holds its own on all audible parameters, trailing only slightly on bass authority.
Ah perhaps, but many threads conclude that the smyasym performs much better when combined with a good pre-amp. There is a Naim nac copy on ebay which sounds pretty good value. I wonder if this could be the missing link!
The NAC clone looks OK - it's all discrete, which should be nice for tweaking and modding. However, an op-amp pre based on the LM4562 or OPA2134 should be able to trump just about any discrete design. I have a locally-made "Triumph" board which could be just the ticket - I'll reverse-engineer the schematic and post it here in due course. It's very simple and uses a single quad op-amp, which is perfect for an LME49740 upgrade.
Meanwhile, the Symasym/Goldmund is sort of growing on me. Maybe it will improve as the electrolytics break in - it certainly sounds a bit more airy and open now than a few hours ago, but it could be just the music I'm listening to now (The Essential Billy Joel) having been mastered better.
Meanwhile, the Symasym/Goldmund is sort of growing on me. Maybe it will improve as the electrolytics break in - it certainly sounds a bit more airy and open now than a few hours ago, but it could be just the music I'm listening to now (The Essential Billy Joel) having been mastered better.
Last edited:
Another cute thing (actually the cutest thing) about the Goldmund/Symasym - the little Omron relay in the transparent plastic housing that clicks on like clockwork about 4 seconds after the power button is pressed. I know, it's just a stock upc1237 speaker protection circuit - but I'm a sucker for precision electromechanical gadgets, and this fits the bill just fine. Maybe they should also put a Nixie or something that counts down the seconds until the relay clicks on the next rev of the board...
Do you think this sort of thing is better,
unbalance input change balance output preamplifier kit on eBay (end time 27-Apr-10 04:00:03 BST)
I agree with the relay. I have to watch it click on like magic.
unbalance input change balance output preamplifier kit on eBay (end time 27-Apr-10 04:00:03 BST)
I agree with the relay. I have to watch it click on like magic.
OK, I simulated the Symasym/Goldmund on LTSpice to try to understand what was going on. With the values as supplied in the kit, the phase margin at unity loop gain was only about 25-30 degrees, i.e. it was fairly closely to oscillating at around 2 MHz. That could account for the output transistors running fairly hot even at moderate Iq of 100 mA.
I played around with the feedback resistor values in the sim to try to increase the phase margin. With 27k/680 (instead of 22k/1k) substituted in the feedback network, the phase margin improves by almost 50 degrees (to ~80 degrees). The closed loop gain is now ~41, so substitution gives up about 6 dB of loop gain, and the THD20 is also correspondingly higher. However, most of the additional distortion is H2, so it doesn't hurt the sonics that much. The additional phase margin however makes up for the increased distortion, and some of the graininess is reduced.
To summarize the mod: Replace the feedback resistor network - change 22k/1k to 27k/680 ohms. I used Allen-Bradley Carbon Composition resistors for the 27k, but that's entirely up to the taste of the builder.
First impressions of audible sonics: Not much negative impact. Graininess reduces, maybe some sibilance and ringing is reduced. Bass is less tight and focussed, but that's subtle. Overall audible sonics improves across the board despite higher distortion.
I played around with the feedback resistor values in the sim to try to increase the phase margin. With 27k/680 (instead of 22k/1k) substituted in the feedback network, the phase margin improves by almost 50 degrees (to ~80 degrees). The closed loop gain is now ~41, so substitution gives up about 6 dB of loop gain, and the THD20 is also correspondingly higher. However, most of the additional distortion is H2, so it doesn't hurt the sonics that much. The additional phase margin however makes up for the increased distortion, and some of the graininess is reduced.
To summarize the mod: Replace the feedback resistor network - change 22k/1k to 27k/680 ohms. I used Allen-Bradley Carbon Composition resistors for the 27k, but that's entirely up to the taste of the builder.
First impressions of audible sonics: Not much negative impact. Graininess reduces, maybe some sibilance and ringing is reduced. Bass is less tight and focussed, but that's subtle. Overall audible sonics improves across the board despite higher distortion.
Do you think this sort of thing is better,
unbalance input change balance output preamplifier kit on eBay (end time 27-Apr-10 04:00:03 BST)
Yup, it looks well laid-out with lots of space, and sockets for the opamps to allow rolling. It's a bit expensive at ~$32 shipped. Maybe it would be a better deal if available as a bare PCB. With a pair of LM4562s instead of the 5532s, it should be very clean sounding.
Thanks, I will get on with the feedback mod. Sound like a good idea, and as a bonus it should run a tiny bit cooler. I will get the pre-amp. I'm too lazy to buy all the seperate bits. Iv'e got a couple of LM4562s to replace the 5532s. I used a ladder type dact pot for volume on the Yuanjing amp, but I don't think they are worth the extra cost. I will stick with a simple cheap alps for this one I think. RS have not got the case for a few weeks, so there will be no hurry with this one. Please let me know if you find any more mods worth doing. I noticed in my kit there are two 22pf caps, but in the sellers picture he has got a couple of micas there instead. I wonder what was supplied in yours?
Please let me know if you find any more mods worth doing. I noticed in my kit there are two 22pf caps, but in the sellers picture he has got a couple of micas there instead. I wonder what was supplied in yours?
A possible future mod, still in simulation for now: Replace the 2sk170 FETs in the LTP with high-gain small-signal NPN BJTs like the 2sc2240, 2sc1845, bc550c, or 2sc3112. At first glance it gives up some phase margin, but they can also be run with higher tail currents and lower emitter degeneration resistors - giving a net benefit of up to 10..20 dB lower distortion. This idea works well with super-beta NPN transistors for the LTP because of the cascoded LTP. I'm not doing this mod on the first board, but I have a second board on its way on which I may try it if the simulations look OK.
I got 2 x 22 pF ceramic (yellow), 2 x 47 pF ceramic (yellow) and 2 x79 pF silver-mica (red-brown 500V Saha). The 79 pF silver micas came inserted (but not soldered) in the board at the locations marked 47 pF. The 22 pF were Miller caps for the PNP VAS transistors - linearity is important here, so I substituted them with 22 pF, 300V silver-mica sourced locally (also Saha, I think). I didn't use ceramics in the signal path anywhere except for the RF suppression caps at the FET inputs, where the voltage swing is not large and the non-linearity of the ceramics doesn't contribute much to distortion.
At the 22 pF Miller cap location, it's ok to use anything from 22 pF to 68 pF. In the simulations, higher values cause more 3rd harmonic (and higher harmonic) distortion, but it also contributes to phase margin and stability. Something like 27 or 33 pF should be a good compromise, but the marked 22 pF also works fine. However, replacing it with a styroflex or silver-mica will improve linearity and audible sonics.
Last edited:
Thanks, Iv'e now changed the 22pf for 27pf silver mica and replaced the feedback resistors with 27k and 680 ohms. I see there is a similar board to the Yuanjing out there with plug in pre-amps. I'll see if I can find it again and post it here for anyone interested.
The Symasym/Goldmund is opening out nicely - it probably takes a while for the 22,000 uF TS-UP caps to burn in, because of the large electrode foil area that has to be formed. It has probably done about 10 hours now, and the detail and imaging are improving steadily.
I made one more mod - replaced the 220uF/50V Panasonic FM Ci DC-blocking caps in the NFB network with 180uF/16V Nippon Chemicon PS Polymer caps. The audible impact is subtle, but it's probably smoother and has darker silences now, i.e. more headroom. The FMs were already very good, so the audible impact of the polymers is small. Probably the main advantage of the polymers is that there's no electrolyte and hence no electrode forming and no burn-in time.
I made one more mod - replaced the 220uF/50V Panasonic FM Ci DC-blocking caps in the NFB network with 180uF/16V Nippon Chemicon PS Polymer caps. The audible impact is subtle, but it's probably smoother and has darker silences now, i.e. more headroom. The FMs were already very good, so the audible impact of the polymers is small. Probably the main advantage of the polymers is that there's no electrolyte and hence no electrode forming and no burn-in time.
I got my 3X 3886 Yuanjing amp board in the mail Saturday. looks reasonably well built. haven't powered it up yet. the input ground is the same as the power and speaker ground plane so i will try that 10 ohm resistor mod. board looks reasonable. plenty of room to mod. the input DC cap area on the board has multiple solder holes for various sizes of capacitors so that could be changed. looks to have some sort of film cap stock Jie Deng brand. I intend to use this for a bass guitar practice amp so i wont be as critical about it. but seems like a good price if it works.
Zc
Zc
Attachments
A friend of mine bought a pair of these Yuanjing 3X boards for a high power stereo and he is very pleased with them. A question for Linuxguru. The 27k and 680ohm resistors I fitted were vishay 1W metal film. Are these OK. They are all I had in my cupboard.
The 27k and 680ohm resistors I fitted were vishay 1W metal film. Are these OK.
Those should be fine, if you can fit them in the space provided - it's a tight fit. Also, be careful when desoldering components on this board - the copper looks like it might delaminate easily with excess heat, unlike the red Yuanjing boards, which are very sturdy in this respect.
This will probably limit some of the experimental mods on this board - I'll await the second board before attempting the more ambitious mods.
- Status
- Not open for further replies.
- Home
- Amplifiers
- Chip Amps
- "Yuanjing" Gainclone 3886 - eBay amazing value ?