Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
question: what voltage do we want on the front end power rails ?
I can readily adjust this by choosing the resistor value in the RC rail filter.
My power supply is still not built, the transformer is not to hand but I believe I'll end up with +/42V rails when done with around +/-40V on the front end with typical RC filter values. That will put a lot of thermal stress on the VAS - spice says 320mW. It does preserve head-room of course. The transistor is rated to 1,000mW maximum in ambient so it would appear there is enough derating. Still, I believe the NAP 160 ran with +/-34V rails. I can easily create this voltage for the front-end. Perhaps it is important for preserving the sound signature (even though max power output will be constrained) ?
I can readily adjust this by choosing the resistor value in the RC rail filter.
My power supply is still not built, the transformer is not to hand but I believe I'll end up with +/42V rails when done with around +/-40V on the front end with typical RC filter values. That will put a lot of thermal stress on the VAS - spice says 320mW. It does preserve head-room of course. The transistor is rated to 1,000mW maximum in ambient so it would appear there is enough derating. Still, I believe the NAP 160 ran with +/-34V rails. I can easily create this voltage for the front-end. Perhaps it is important for preserving the sound signature (even though max power output will be constrained) ?
You can't be blamed for wanting lower noise levels by adding RC filtering to the front end but referring to the common NAP 250 schematic, it has regulated 40V rails as shown and no RC rail filtering. There will be a difference in sound with filtering for a number of reasons but I can't say how much as I have never tried it on the real thing.
NAP140 is essentially the same but has unregulated 34V rails. To keep some level of parity, why not use a 25V+25VAC transformer allowing 35V rails. Alternatively, go the whole hog with the 30 +30VAC transformer by adding regulation for 34V rails so that the supplies to the front end and output stage can stay connected.
'Only my view, but I think the common rails should stay or you'll have something even less like the original than present Ebay clones - from a design POV. I suspect the sound will actually be noticeably different .
NAP140 is essentially the same but has unregulated 34V rails. To keep some level of parity, why not use a 25V+25VAC transformer allowing 35V rails. Alternatively, go the whole hog with the 30 +30VAC transformer by adding regulation for 34V rails so that the supplies to the front end and output stage can stay connected.
'Only my view, but I think the common rails should stay or you'll have something even less like the original than present Ebay clones - from a design POV. I suspect the sound will actually be noticeably different .
As far as I know RF into amplifers causes a DC shift that looks like uncorrected crossover distortion ( base emitter acts as detector ). It looks to be the long tail pair in doubt. We often filter the +ve input, the output choke I guess helps. As RF would be a common mode problem the Zobel won't do much nor the input filter I guess? We often say the input filter is for RF, I have doubts about that. Being that ground is low impedance to actual ground perhaps the common mode problem becomes differential mode by happy accident.
Member
Joined 2009
Paid Member
Output Choke
In response to Bigun etc.
I found I solved a problem many decades ago with my neighbour's son putting out 100W/27MHz or so from 20 feet away definitely was solved with a simple air-cored inductor. However, a while ago I experimented with using a common-mode inductor with just 2 X2 turns on a ferrite toroid. This is wired in the speaker feed and return and appears to work very well with a conventional amplifier and also with my class A balanced output amplifier. Worth a try?
In response to Bigun etc.
I found I solved a problem many decades ago with my neighbour's son putting out 100W/27MHz or so from 20 feet away definitely was solved with a simple air-cored inductor. However, a while ago I experimented with using a common-mode inductor with just 2 X2 turns on a ferrite toroid. This is wired in the speaker feed and return and appears to work very well with a conventional amplifier and also with my class A balanced output amplifier. Worth a try?
Good time! Here is a message received from a friend who made a clone, with "incorrect polarity of electrolytic capacitors of 68-100 μF"
"Recently I took this capacitor to take measurements, and in one channel the tantalum was already dead. It was connected according to the scheme - "plus" to the transistor ".
Be careful when using these capacitors!
"Recently I took this capacitor to take measurements, and in one channel the tantalum was already dead. It was connected according to the scheme - "plus" to the transistor ".
Be careful when using these capacitors!
I last heard the chrome bumper 250s and 135s. What the newer models sound like I can't say although there are some mixed options on Pinkfish. I would say the 250 is the quintessential Naim sound.
Maybe some discussion would be helpful about what the "Naim sound" is?
My experience of the chrome bumper 250:
- Overall relaxed (natural) and with clear and snappy midrange. Particularly realistic with acoustic instruments, guitars, brass, close-miked vocals.
- Properly tuneful bass and PRaT
- Lively and fun but not fatiguing
- Quite good emotional communication...almost as good as valves.
- Unconstrained dynamics. Doesn't sound like it is running out of oomph. Loud percussive sounds will rip your head off but somehow you feel better for it.
- not as good as the big American amps for soundstage
- not quite as stunningly natural as I prefer, a little closed in.
A home brew copy is unlikely to achieve all of that but if you are lucky you may get some of it and we can discuss tweaks later. It will also depend heavily on your source. What is your front end?
I agree with Ian's recommendations to regulate the supply upstream albeit it a PITA. Keep your caps on the pcb too, tho. Upstream regulation will reduce noise. I would keep the input stage LPFs but ditch the diodes, because your circuit will have worse psu noise than a Naim.
Member
Joined 2009
Paid Member
I'd forgotten about those guys, I bought my last hi-fi from Soundstage. The chap who runs it is very good.
So!
1) As the regulated NAP250 has the naim sound then we can say that a regulated power supply is not going to rob the amplifier of it's sound.
2) Hence, we can say that the front end power supply rails need not be connected to the output stage power rails directly because in a fully regulated design like the NAP250 there won't be any signal-related stuff on the rails to allow 'communication' between output stage and front-end.
Conclusion: we can regulate only the front-end only and still have the Naim sound. The output stage has pretty good PSRR all by itself. I'm not saying regulation doesn't help but regulation of the output stage is beyond the scope of this project.
Make sense ?
And that still leaves my question: drop the voltage at the front-end to +/-34V (tames power dissipation in the Zetex devices a little and is closer to the original design) or not ?
gholl - good to know !
So!
1) As the regulated NAP250 has the naim sound then we can say that a regulated power supply is not going to rob the amplifier of it's sound.
2) Hence, we can say that the front end power supply rails need not be connected to the output stage power rails directly because in a fully regulated design like the NAP250 there won't be any signal-related stuff on the rails to allow 'communication' between output stage and front-end.
Conclusion: we can regulate only the front-end only and still have the Naim sound. The output stage has pretty good PSRR all by itself. I'm not saying regulation doesn't help but regulation of the output stage is beyond the scope of this project.
Make sense ?
And that still leaves my question: drop the voltage at the front-end to +/-34V (tames power dissipation in the Zetex devices a little and is closer to the original design) or not ?
gholl - good to know !
Last edited:
I think RC to regulate rail for LTP and VAS is okay if no more than 4v voltage drop. Highet than that the change is too much (assuming the cct already tuned up).
Bigun, I have a stabilized voltage, like in the NAP250. The voltage level is +/- 32V. This was done in my first clone, and so did in the next few. Sound with stabilization is slightly better than when it is not available. And, at this voltage level - there are no problems with heating the transistors (any).
And, according to "naim sound" - it will be determined more by NAC-pre amp.
Yes, and the separation chain (RC) between the cascades, on the power circuit did not use.
And, according to "naim sound" - it will be determined more by NAC-pre amp.
Yes, and the separation chain (RC) between the cascades, on the power circuit did not use.
Member
Joined 2009
Paid Member
Well I have it powered up using a current-controlled bench supply. The supply is limited to +/-20V but it has some 5v taps so I've put them in series and created a current-controlled +/-25V bench top supply.
No smoke
It's just soaking on the bench for now to allow time for the heatsink to warm up.
Without adjusting any pots the idle current seems to be around 20mV across 0R22 which is 90mV. The power supply current meter shows 100mA per rail so there's around 10mA going through the front-end which makes good sense.
It'll take quite awhile to warm up the heatsink with only 2W dissipation per output device but I want to see if the thermal compensation is even close.
fyi - The bench top power supply voltage is too low to 'turn on' the speaker output relay. My circuit is designed to open the relay when the power rails to the output stage dip below roughly 34V (set by a zener so it can be easily changed). This is to ensure rapid disconnect on power-down.
No smoke
It's just soaking on the bench for now to allow time for the heatsink to warm up.
Without adjusting any pots the idle current seems to be around 20mV across 0R22 which is 90mV. The power supply current meter shows 100mA per rail so there's around 10mA going through the front-end which makes good sense.
It'll take quite awhile to warm up the heatsink with only 2W dissipation per output device but I want to see if the thermal compensation is even close.
fyi - The bench top power supply voltage is too low to 'turn on' the speaker output relay. My circuit is designed to open the relay when the power rails to the output stage dip below roughly 34V (set by a zener so it can be easily changed). This is to ensure rapid disconnect on power-down.
Congratulations on the early success Gareth - given some more spare time, it won't long before you're on the home stretch and your own audition
BTW, rensli tried simply regulating the front end supply of his 140 clone and didn't like it at all, though I'm unsure of all his implementation details - only the proposed schematic with 3 terminal regs and a separate power supply for them. I guess there's a little more to explain there.
BTW, rensli tried simply regulating the front end supply of his 140 clone and didn't like it at all, though I'm unsure of all his implementation details - only the proposed schematic with 3 terminal regs and a separate power supply for them. I guess there's a little more to explain there.
Last edited:
Member
Joined 2009
Paid Member
I'm not happy with the bias and thermal compensation yet so this will have to be addressed before the o'scope gets poweres up. After an hour of soaking the bias was stable. I wired up a +/-34V supply.
BUT
a) I discovered the bias is sensitive to me blowing air on the amplifier. I've traced this to the CCS transistor. Spice simulations confirm it's relatively high sensitivity and impact on output bias current. I believe it's OK once sealed in a case but I don't like it. I am planning to place it in physical contact with the two diodes that are supposed to ensure better temperature behaviour of the CCS
Also
b) I fitted the wrong value of a resistor in the bias network that limits the range of adjustment. It'll need to be replaced.
BUT
a) I discovered the bias is sensitive to me blowing air on the amplifier. I've traced this to the CCS transistor. Spice simulations confirm it's relatively high sensitivity and impact on output bias current. I believe it's OK once sealed in a case but I don't like it. I am planning to place it in physical contact with the two diodes that are supposed to ensure better temperature behaviour of the CCS
Also
b) I fitted the wrong value of a resistor in the bias network that limits the range of adjustment. It'll need to be replaced.
Member
Joined 2009
Paid Member
Much better, bias is stable now. I've set it for 5mV across Re giving 23mA (+/-5%) of bias current through the output. Final version can be a bit higher.
This is about right if you believe these results: Bias current and the Naim power amp - pink fish media
This is about right if you believe these results: Bias current and the Naim power amp - pink fish media
Member
Joined 2009
Paid Member
crappy basement set-up !
I hooked up the 'scope. Used the probe compensation test point as a signal source, gives 0.5Vpp square wave at 1kHz. It's much cleaner than my signal generator so it's my first go-to signal.
Without a load the output is clean and shows an amplifier gain of roughly x26. On first connecting it the green LED goes out and comes back on - this square wave gets the speaker protection all excited.
With a 7R5 resistive load we start to see a tiny bit of overshoot on the +ve side. I'm not using a scope probe here, just a wire into the scope input.
Of course, I had to have a quick listen. All I had to hand was a floorstander DIY speaker which I stuck on the table top and my Blackberry as a source playing Youtube. It made music (but let's not judge based on my BB - it wasn't that great).
I hooked up the 'scope. Used the probe compensation test point as a signal source, gives 0.5Vpp square wave at 1kHz. It's much cleaner than my signal generator so it's my first go-to signal.
Without a load the output is clean and shows an amplifier gain of roughly x26. On first connecting it the green LED goes out and comes back on - this square wave gets the speaker protection all excited.
With a 7R5 resistive load we start to see a tiny bit of overshoot on the +ve side. I'm not using a scope probe here, just a wire into the scope input.
Of course, I had to have a quick listen. All I had to hand was a floorstander DIY speaker which I stuck on the table top and my Blackberry as a source playing Youtube. It made music (but let's not judge based on my BB - it wasn't that great).
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.