John,
thanks for that, I tend to think in pictures so that helps
I'll probably 'sim' it over the next few days - I'm off to the audio jumble tomorrow down in Tonbridge, and for Monday, SWMBO has added a couple of house items to the 'to do' list.
The audio jumble is a good place for retro stuff and I ned some 100uF+100uF caps for my old Radford LDO. Great for memories as well, but I see they are pushing up the prices again to get in at a decent time.
Kind regards
Mike
thanks for that, I tend to think in pictures so that helps
I'll probably 'sim' it over the next few days - I'm off to the audio jumble tomorrow down in Tonbridge, and for Monday, SWMBO has added a couple of house items to the 'to do' list.
The audio jumble is a good place for retro stuff and I ned some 100uF+100uF caps for my old Radford LDO. Great for memories as well, but I see they are pushing up the prices again to get in at a decent time.
Kind regards
Mike
Looks like you have really been working hard John!
Just to put things in perspective, if you compare it to the RC circuit I added to Mike's original circuit, what do you think? Is there something fundamentally wrong with it? If so, don't be too polite, just say it. I'm a novice, but I can handle the truth!
I found it as a simple and effective solution, it shares the currents nicely, and seems to have high efficiency on voltage swing and current use. It also had pretty good simulated distortion performance just as the original from Mike. However I suspect the uneven currents in Mike's original design might add some extra distortion when built.
I can also see that with the RC it only needs abt 8,5A Iq to acheive full output swing, and without it it needs abt 12A.
Increasing the current sharing resistors will create more heat (on PCB?) and decrease efficiency of the circuit? Ok, maybe class A and efficiency is a stupid subject.. sorry about that!
Last edited:
I missed that post. Yes it looks an interesting idea, but goes back to bootstrap days. Not that that is a big issue when everything else is AC coupled.
Not sure how you control the quiescent current though. I suspect you will still need to use a local feedback resistor. Nor how you control the current sharing. Simulators using the same model for parallel transistors don't understand manufacturing differences - that needs additional models.
Another possible mod might be to wire all the bases of the MJ15004's together (only in each half, obvs.) so as to utilise your existing emitter resistors. But then you will still need smaller resistors in the emitters of the drivers to ensure the current is balanced in those as the mod. only shares the output transistor currents.
The emitter resistors of 0.68 ohms I tried are larger than needed for current sharing, so there is scope to reduce those (and distortion).
Need to simulate at different temperatures - especially with the output devices heating up.
That will probably need a different model to swap in for the MJ15004's with a different temperature setting compared with the rest of the amp.
Not sure how you control the quiescent current though. I suspect you will still need to use a local feedback resistor. Nor how you control the current sharing. Simulators using the same model for parallel transistors don't understand manufacturing differences - that needs additional models.
Another possible mod might be to wire all the bases of the MJ15004's together (only in each half, obvs.) so as to utilise your existing emitter resistors. But then you will still need smaller resistors in the emitters of the drivers to ensure the current is balanced in those as the mod. only shares the output transistor currents.
The emitter resistors of 0.68 ohms I tried are larger than needed for current sharing, so there is scope to reduce those (and distortion).
Need to simulate at different temperatures - especially with the output devices heating up.
That will probably need a different model to swap in for the MJ15004's with a different temperature setting compared with the rest of the amp.
As I understand it, Iq is still from the ccs (capacitor decouples the DC), the capacitor and 2k2 resistor takes care of impedance matching the AC signal for the upper bases to do the current sharing when signal is applied. It looks to be working all right in the sim.. + I like it since it has similarities to the original JLH
I quickly tried it on a 96 JLH sim too, but it did not work well. Maybe needs some tuning.
I quickly tried it on a 96 JLH sim too, but it did not work well. Maybe needs some tuning.
The worst thing with Sims is they can force your thinking down the wrong road. In other words the defect is the flavour of the design.
On a similar track I now feel my worthy JLH preamp is ok. The NAD 3020 plus unique Riaa stage. I had guessed the distortion of the stage to be 0.003% x 100 the gain. Well MC33078 publishes a curve for that at 0.05%. That's incredible and is up to 10 kHz. The MC 33078 is cheap and really nice sounding. Even at gain 1000 it's 0.13%. I have reduced the gain to 65 which is about the same as my phone in level. Volume control about two o'clock to get .5 watts. I increased the arm mass and extended the preamp bass to 2.5 Hz -3dB. The sound now is as I imagined it could be. Classical music works rather better than I have heard it before. The Hadcock is a fine arm and the Denon if correctly matched to the preamp stunning.I have a Lyra Helikon. I am not sure it will be better. There is a MC33077 which seems related. It could almost be used open loop. A most unusual opamp, never heard one. On paper it looks formidable. It's even low cost. I have no Idea if it sounds good.
On a similar track I now feel my worthy JLH preamp is ok. The NAD 3020 plus unique Riaa stage. I had guessed the distortion of the stage to be 0.003% x 100 the gain. Well MC33078 publishes a curve for that at 0.05%. That's incredible and is up to 10 kHz. The MC 33078 is cheap and really nice sounding. Even at gain 1000 it's 0.13%. I have reduced the gain to 65 which is about the same as my phone in level. Volume control about two o'clock to get .5 watts. I increased the arm mass and extended the preamp bass to 2.5 Hz -3dB. The sound now is as I imagined it could be. Classical music works rather better than I have heard it before. The Hadcock is a fine arm and the Denon if correctly matched to the preamp stunning.I have a Lyra Helikon. I am not sure it will be better. There is a MC33077 which seems related. It could almost be used open loop. A most unusual opamp, never heard one. On paper it looks formidable. It's even low cost. I have no Idea if it sounds good.
I have just said classical music sounds better. Example. Early digital recordings sound far better on LP rather than CD players of the period. Karajan the Planets is different to Boult or Previn. It's nice in it's own way. The only defect is the room reverberation is missing. Analogue recordings of the same venue are alive with reverberation. As Karajan was typical of his generation he dictated that the microphone placement didn't interfere with his eye line and was not changed.The system makes it easy to hear that yet enjoy early digital. I have tons of it mostly bought from charity shops. The reason early digital sounds better on LP is it's direct from the master tape. I had a long conversation with Peter Andre of EMI. He said it's because modern sound engineers never do the same examinations as a conductor that he did means that they have no idea what is a good sound. I agree.
I think Sergeant was shot at in Egypt. Beecham apparently said " Funny I never thought of the Egyptians as a musical people" He called him flash harry.
The boss of Thorens Mr Leitner was a concert pianist who steadfastly never played piano. His father Ferdinand Leitner a famous conductor. I really liked Leitner. I might be one of the few who did. He looked like Karajan. He introduced me to the Thorens importer to the USA before the war! A famous diva who's name I forget also. Leitner was the real deal.
The boss of Thorens Mr Leitner was a concert pianist who steadfastly never played piano. His father Ferdinand Leitner a famous conductor. I really liked Leitner. I might be one of the few who did. He looked like Karajan. He introduced me to the Thorens importer to the USA before the war! A famous diva who's name I forget also. Leitner was the real deal.
I had this PNP JLH apart the other day to replace the bootstrap cap with a larger one, so I updated the schematic with that, and also found an error: I have no feedback cap in the amp, so I removed it from the schematic too.
I can't say I can hear any difference with the lager bootstrap cap, but I do love the sound from this amp! I noticed a significant improvement in sound when I switched back to this one from another JLH.
I can't say I can hear any difference with the lager bootstrap cap, but I do love the sound from this amp! I noticed a significant improvement in sound when I switched back to this one from another JLH.
Attachments
Have you simulated the transient response (low frequency) of the floating ground arrangement?
I also had a design once (class AB) using the floating ground ... until I simulated it.
I found that decoupling the supply lines with a large capacitor helped. That means adding a 10k equivalent to R16 in the - line and something like 1mF each to ground.
I also had a design once (class AB) using the floating ground ... until I simulated it.
I found that decoupling the supply lines with a large capacitor helped. That means adding a 10k equivalent to R16 in the - line and something like 1mF each to ground.
Member
Joined 2009
Paid Member
I've simulated it before and it requires a large amount of capacitance and it suffers from the same risk of low frequency instability as a single output capacitor - as you would expect.
fyi - slightly OT - What I did on a Class AB amplifier which employed this floating ground was to connect the floating ground to the centre tap of the power transformer secondary through a power resistor. This is somewhat akin to split rail supply and hence it 'anchors' the ground at DC but it protects the speaker from excessive DC current in the case of a fault. This protection only works for mono construction in which additional channels have separate exclusive floating grounds.
fyi - slightly OT - What I did on a Class AB amplifier which employed this floating ground was to connect the floating ground to the centre tap of the power transformer secondary through a power resistor. This is somewhat akin to split rail supply and hence it 'anchors' the ground at DC but it protects the speaker from excessive DC current in the case of a fault. This protection only works for mono construction in which additional channels have separate exclusive floating grounds.
It seems I was a bit hasty in judging this amp.
I now have it sounding very good.To get it sounding right I had to try a few different speakers to find the right ones and also experiment with different cables.It seems to be one of those amps that is cable and speaker fussy.Which is fair enough and probably just shows it is transparent and revealing.
The speakers which combine best are Yamaha NS75Ts which are fabulous sounding things [like Yamaha NS1000s but much better in every way] but they are ultra revealing and ultra high resolution with a tendency to sound a bit sharp.They sound great with the Chinese Hood amp.They are an easy load.My guess is that is probably the key to hearing these amps at their best.Which I should have considered from the start.Luckily I have a couple of speakers that are an easy load.Most speakers are not.
Last edited:
You should try to avoid low impedance and low efficiency. Low impedance will raise distortion, and low efficiency limits the playback level.
Give it some time, and you might even want to start tweaking it
Yes.. and seen it on a scope too, I mentioned it some time ago in this thread. There is a visible (on woofers) LF oscillation around 1Hz at startup, but it dies after abt 2 cycles. It can also be made to 'swing' a bit from steady state with correctly spaced tone bursts, I think I posted a scope picture of this. I have chosen to ignore it, since the amp just sounds so nice, but it is in the back of my head..
I have also made a spice model of the PNP amp that is almost spot on in distortion simulations compared to the real thing, even if the low power transistors are not the same as the real thing. It seems getting the models for the 2SA1216 outputs made the difference.
After listening some more to it after I changed the bootstrap cap, I have a feeling the sound has changed a bit. Not only LF, but mid/treble too, and that should be imagination. Anyway, I'm thinking about getting some 'audio' caps to try in this position..
Last edited: