Before I go shopping for components does any of this alter if the Amp is the EL34-B model not the FU50?
Last edited by a moderator:
Basically this is what I ended up with and am very happy with it
Trying to make sense of the Sweet Peach
It's a significant mod though, involving a new top plate for the smaller tubes.
It's also controversial as it dispenses with the ridiculous obsession of including the phasey band limited OPT within a huge global feedback loop which IMO is the absolute last place an OPT should be.
The problem with the OPT inside the GNFB loop is the areas the need the most help are the bass and treble, which are also the OPTs points of highest attenuation, so those areas get the least feedback. IMO mild feedback is worse than no feedback. Additionally treble feedback around an item with phase delay is questionable for HF stability and therefore sonics.
So the whole concept of the change is to put a driver/pentode pair as a high gain high feedback OPT driver to drive the OPT with a low impedance and a correct waveform, and then let the OPT deal with the output itself.
This ensures fast, heavy NFB at the treble, mid and bass to _drive_ the transformer correctly, which then for my ears sounded a lot better.
The nice thing about SE amps if they're dead easy to mod, unlike push-pull which has far more components and less modding options IMO.
Trying to make sense of the Sweet Peach
It's a significant mod though, involving a new top plate for the smaller tubes.
It's also controversial as it dispenses with the ridiculous obsession of including the phasey band limited OPT within a huge global feedback loop which IMO is the absolute last place an OPT should be.
The problem with the OPT inside the GNFB loop is the areas the need the most help are the bass and treble, which are also the OPTs points of highest attenuation, so those areas get the least feedback. IMO mild feedback is worse than no feedback. Additionally treble feedback around an item with phase delay is questionable for HF stability and therefore sonics.
So the whole concept of the change is to put a driver/pentode pair as a high gain high feedback OPT driver to drive the OPT with a low impedance and a correct waveform, and then let the OPT deal with the output itself.
This ensures fast, heavy NFB at the treble, mid and bass to _drive_ the transformer correctly, which then for my ears sounded a lot better.
The nice thing about SE amps if they're dead easy to mod, unlike push-pull which has far more components and less modding options IMO.
Just thought I'd update this with a few slight details as I had to search my memory (and then look at the tubes) recently - and it's not very clear.
The amp is still in use although the volume pot is noisy now so I may pull out the headphone socket, old pot and put a new remote control pot in. It still has a very large airy, effortless sound with a lovely sweet but precise treble balanced by some thumping, solid bass with proper super low bass notes in there too.
The sweet peach is probably limited by the OPTs but it doesn't show, it would be interesting to hear the circuit with top notch OPTs though. It's a great DIY project as being S.E. its far simpler than a P-P amp.
You can still get a slightly later model here it seems
Aliexpress.com : Buy XiangSheng Sweet Peach SP FU50 Tube Amplifier HIFI EXQUIS FU50 Signal ended MM Phono Headphone output USB Decode mp from Reliable amplifier pre amp suppliers on HIFI EXQUIS
Tubes:
The tube lineup I ended up with was ECC88 (Amperex Bugleboy) input, Russian 2 x 6N2P-EV (for the SRPP) and Russian GU50 output tubes (Idling at 55mA, measuring 0.55V over the 10ohm resistors from pin1 on the GU50 bases).
SRPP precautions:
The upper SRPP triode for both channels are in the 6N2P-EV tube on the RHS powered by a floating 6V supply (Tiny potted transformer via a 4R7 ohm ceramic resistor) so as not to exceed the filament/cathode voltage. I used SRPP as it's easier than finding an anode resistor or making a current source for me, but any high gain driver would do especially as the GU50 is in Pentode mode so should be easy to drive IIRC.
The lower SRPP for both channels is on the LHS tube and uses the old heater supply as does the input tube.
All capacitors were replaced for better ones, and it runs no input capacitors as I figure most stuff has output capacitors anyway.
Given the unusual feedback topology, capless inputs and Russian tubes it's actually rather a subversive amplifier...
Finally got The Peach into pentode mode on the GU50s.
This involves driving a regulated G2 voltage of around 255V, which I tried before with some VR tubes - but every time I loaded them even a tiny bit they started oscillating - anything from 500Hz to 50kHz, which is easy to spot if you have a scope - but tricky without. Now I test everything with a scope and now it's very clean.
To stop the VR tube oscillations I used one of Wavebournes circuit ideas and put an IRF730 source follower in. It doesn't even need a heatsink and sits on a tiny bit of veroboard (the tab is live).
If you just connect up G2 you turn off the tube as well, as the usual 70V bias now changes to about 20V or so, so the peach needs the two earth side resistors on the bias circuit shorting out so one can re-bias.
With the current circuit B+ is at 420V, GU50 bias is at 60mA and there is only feedback between the GU50 and SRPP driver, plus local, no cathode capacitors are used. I still have a few electro caps to bypass but already it is showing massive promise. Pentode mode sounds better too, and I think this is because:
1) There is feedback around the pentode so output impedance is still low
2) The driver has to swing less volts, which makes it more linear
3) After drilling holes for VR tubes, getting the IRF730 mosfet etc I really want it to
The only mods to do now are:
1) A little circuit tidying
2) Protection for the VR tubes/G2 voltage
3) Cap bypassing
4) Perhaps split the middle SRPP resistor to take output from the middle
5) Reference G2 to the cathodes, not ground
6) Try the 6N1p-ev, 6N2p-ev and 6N23p-ev in place of the 6N6p as SRPP tubes
The amp is still in use although the volume pot is noisy now so I may pull out the headphone socket, old pot and put a new remote control pot in. It still has a very large airy, effortless sound with a lovely sweet but precise treble balanced by some thumping, solid bass with proper super low bass notes in there too.
The sweet peach is probably limited by the OPTs but it doesn't show, it would be interesting to hear the circuit with top notch OPTs though. It's a great DIY project as being S.E. its far simpler than a P-P amp.
You can still get a slightly later model here it seems
Aliexpress.com : Buy XiangSheng Sweet Peach SP FU50 Tube Amplifier HIFI EXQUIS FU50 Signal ended MM Phono Headphone output USB Decode mp from Reliable amplifier pre amp suppliers on HIFI EXQUIS
Tubes:
The tube lineup I ended up with was ECC88 (Amperex Bugleboy) input, Russian 2 x 6N2P-EV (for the SRPP) and Russian GU50 output tubes (Idling at 55mA, measuring 0.55V over the 10ohm resistors from pin1 on the GU50 bases).
SRPP precautions:
The upper SRPP triode for both channels are in the 6N2P-EV tube on the RHS powered by a floating 6V supply (Tiny potted transformer via a 4R7 ohm ceramic resistor) so as not to exceed the filament/cathode voltage. I used SRPP as it's easier than finding an anode resistor or making a current source for me, but any high gain driver would do especially as the GU50 is in Pentode mode so should be easy to drive IIRC.
The lower SRPP for both channels is on the LHS tube and uses the old heater supply as does the input tube.
All capacitors were replaced for better ones, and it runs no input capacitors as I figure most stuff has output capacitors anyway.
Given the unusual feedback topology, capless inputs and Russian tubes it's actually rather a subversive amplifier...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.