Bargain beaut - headphone amp
When I found this on Taobao a few days ago I could not believe the price wasn't a mistake, or that it was just for the case with nothing inside. But it turned out to be real, so I ordered one - it arrived just now so I'm taking it apart before having a listen. The case oozes quality and the volume control feels silky smooth.
Here - À³Âü(lehmann) ¶ú·Å À³Âü¹¹¼ÜµÄºÃÉù¶ú·Å£¬È«ÂÁ»úÏ䣬һÌå¶ú·Å-ÌÔ±¦Íø
When I've had a listen (gotta search for my 1/4" adapter) I'll get into modding.... Incidentally for those on 110V, it does have a mains voltage selector switch. Amazing.
I traced out the circuit (still haven't listened) and found its pretty much as shown on the Taobao page. 78/79 15V regulators feed BD139/140s and a TL072 sits between the volume pot and the discrete OPS. This stage resembles a diamond, but instead of the more familiar current sources to the rails, it has 1500ohm resistors. LTSpicing this arrangement (schematic shown) reveals its PSRR is very poor indeed so my first mod's going to be restoring a more 'canonical' diamond output buffer with 2 transistor CCSs.
Update1 - got the CCSs in, biassing all checks out just as in LTSpice, which is very encouraging. So I have a first listen and the dynamics are really, really sucky. A quick check against the trafo-based amp as a reference reveals how totally sucky it is. I wonder if the frequency response is shot, so check that and no, ruler flat out beyond 20kHz. So I buzz out the components around the TL072 and discover the diamond buffer's not in the feedback loop and the TL072's feedback resistors are 3k3 and 470ohm. The light dawns..... too late now but tomorrow morning I'll try including the buffer within the feedback and see how that sounds.
Update2 - OK now we're in the right ballpark - the HF details are no longer being masked and there's a decent sense of scale. I have 33k and 10k as gain setting Rs and the 33k goes to the buffer's output so the opamp's not loaded by it. Not quite the ultimate slam of the trafo amp so I'll be looking at the PSU noise levels but satisfying to listen to.
Later I noticed the lack of slam stays even at very low volume levels so I figured its not due to load induced PSU noise, and checked the grounding. The pot's grounding was on a track on the top of the board, so I've put in my own dedicated ground, also shorted the two input caps as I've driving via a trafo on my DAC, no need for excess caps. This has lowered the noise floor very nicely. Next up fitting some series impedance before the res caps (perhaps CLC) to get residual mains harmonics down.
I've removed the original 4700uF/35V caps and implemented a CLC - 2200uF - 2.2mH - 2 * 2200uF. The inductor's a shielded Sumida type, rated to 400mA, around 2R DC. No fancy caps yet, just normal 'cooking' grade ones. Overall slightly more ease of listening now, but I'll allow time for all the caps to bed in and spin a few disks. At the moment its 'Idomeneo' from a new boxed set of Mozart CDs - the emotion's being conveyed nicely.
Update3 - included the schematic of the CCS'd diamond buffer.
Update4 - more radical overhaul of the design to include output transformers. These are re-purposed 2VA EI cores which have split bobbins. I've rewound both primary and secondary to give 5:1 step down with an output impedance below 2ohms. In order to be able to fit them within the case I took out the 15VA toroid and substituted a 5VA EI core. Let's face it, the 15VA is rather large overkill for such a dinky amp. Having a toroid though is a good idea due to them being less incontinent with the flux than EI. The OPT at the front is a little too close to the revised mains trafo meaning that channel gets a low level 50Hz hum, independent of volume setting. Only audible at very low output levels. I've added more caps to the supply and also an input buffer - the volume control's a 10k one which I consider too low, the classA input buffer (mounted on the reverse of the PCB) raises the Zin to 150k or so. The buffer's just a CCS-loaded EF (BC807-40), AC coupled through X7R caps.
Here - À³Âü(lehmann) ¶ú·Å À³Âü¹¹¼ÜµÄºÃÉù¶ú·Å£¬È«ÂÁ»úÏ䣬һÌå¶ú·Å-ÌÔ±¦Íø
When I've had a listen (gotta search for my 1/4" adapter) I'll get into modding.... Incidentally for those on 110V, it does have a mains voltage selector switch. Amazing.
I traced out the circuit (still haven't listened) and found its pretty much as shown on the Taobao page. 78/79 15V regulators feed BD139/140s and a TL072 sits between the volume pot and the discrete OPS. This stage resembles a diamond, but instead of the more familiar current sources to the rails, it has 1500ohm resistors. LTSpicing this arrangement (schematic shown) reveals its PSRR is very poor indeed so my first mod's going to be restoring a more 'canonical' diamond output buffer with 2 transistor CCSs.
Update1 - got the CCSs in, biassing all checks out just as in LTSpice, which is very encouraging. So I have a first listen and the dynamics are really, really sucky. A quick check against the trafo-based amp as a reference reveals how totally sucky it is. I wonder if the frequency response is shot, so check that and no, ruler flat out beyond 20kHz. So I buzz out the components around the TL072 and discover the diamond buffer's not in the feedback loop and the TL072's feedback resistors are 3k3 and 470ohm. The light dawns..... too late now but tomorrow morning I'll try including the buffer within the feedback and see how that sounds.
Update2 - OK now we're in the right ballpark - the HF details are no longer being masked and there's a decent sense of scale. I have 33k and 10k as gain setting Rs and the 33k goes to the buffer's output so the opamp's not loaded by it. Not quite the ultimate slam of the trafo amp so I'll be looking at the PSU noise levels but satisfying to listen to.
Later I noticed the lack of slam stays even at very low volume levels so I figured its not due to load induced PSU noise, and checked the grounding. The pot's grounding was on a track on the top of the board, so I've put in my own dedicated ground, also shorted the two input caps as I've driving via a trafo on my DAC, no need for excess caps. This has lowered the noise floor very nicely. Next up fitting some series impedance before the res caps (perhaps CLC) to get residual mains harmonics down.
I've removed the original 4700uF/35V caps and implemented a CLC - 2200uF - 2.2mH - 2 * 2200uF. The inductor's a shielded Sumida type, rated to 400mA, around 2R DC. No fancy caps yet, just normal 'cooking' grade ones. Overall slightly more ease of listening now, but I'll allow time for all the caps to bed in and spin a few disks. At the moment its 'Idomeneo' from a new boxed set of Mozart CDs - the emotion's being conveyed nicely.
Update3 - included the schematic of the CCS'd diamond buffer.
Update4 - more radical overhaul of the design to include output transformers. These are re-purposed 2VA EI cores which have split bobbins. I've rewound both primary and secondary to give 5:1 step down with an output impedance below 2ohms. In order to be able to fit them within the case I took out the 15VA toroid and substituted a 5VA EI core. Let's face it, the 15VA is rather large overkill for such a dinky amp. Having a toroid though is a good idea due to them being less incontinent with the flux than EI. The OPT at the front is a little too close to the revised mains trafo meaning that channel gets a low level 50Hz hum, independent of volume setting. Only audible at very low output levels. I've added more caps to the supply and also an input buffer - the volume control's a 10k one which I consider too low, the classA input buffer (mounted on the reverse of the PCB) raises the Zin to 150k or so. The buffer's just a CCS-loaded EF (BC807-40), AC coupled through X7R caps.
Total Comments 2
Comments
-
You can save yourself some trouble and buy the case on ebay with all the components removed already!
Posted 5th June 2015 at 06:12 AM by rjm
-
Excellent - the price equivalent is 204rmb, meaning I got all the internals for 165rmb. I hope its available on Taobao for no higher a price
But actually most of the components I'm keeping, I'll post up a pic sometime.....Posted 5th June 2015 at 01:51 PM by abraxalito
