I was under the impression that you could only drive an output stage to it's supply potential maximums. Any output drive capasity beyond this is completely wasted.
Anyway, there will be nothing to load down the voltage amp supply by virtue of the high impedance load it drives. There is no advantage given by a "springy" supply as it has a mostly constant current demand on it. There is no loss in dynamic headroom. A regulator will isolate the load from outside influences. This is desirable. Are you referring to the output stage power supply perhaps? Then your argument makes sense.
I had a customer once with a Nakamichi CA-7 preamp. One channel had about 108dB S/N, the other was around 122 dB - unweighted. The customer complained about the noise difference and I got rooked into "repairing" the preamp. I solved the problem, both channels now 122 dB. Lead dress was the issue and I needed to install a shield. If the supply was unregulated it would not even approach these figures. The fact the customer heard this was one thing. The other thing was the preamp exceeded it's S/N specs untouched. But the guy heard it. The fault was supply ripple. I am a believer.
Whether you accept this or not is up to you, but it is a fact proven in the field. My designs always seem to have more "air" and dynamics with a regulated front end.
-Chris
Anyway, there will be nothing to load down the voltage amp supply by virtue of the high impedance load it drives. There is no advantage given by a "springy" supply as it has a mostly constant current demand on it. There is no loss in dynamic headroom. A regulator will isolate the load from outside influences. This is desirable. Are you referring to the output stage power supply perhaps? Then your argument makes sense.
I had a customer once with a Nakamichi CA-7 preamp. One channel had about 108dB S/N, the other was around 122 dB - unweighted. The customer complained about the noise difference and I got rooked into "repairing" the preamp. I solved the problem, both channels now 122 dB. Lead dress was the issue and I needed to install a shield. If the supply was unregulated it would not even approach these figures. The fact the customer heard this was one thing. The other thing was the preamp exceeded it's S/N specs untouched. But the guy heard it. The fault was supply ripple. I am a believer.
Whether you accept this or not is up to you, but it is a fact proven in the field. My designs always seem to have more "air" and dynamics with a regulated front end.
-Chris
Way to go, Anatech,
Lots of nice bouncy supply for the output stage so you take advantage of it's transient excess ability, then regulate the early stages to purify the lot. Not only does it make for nice clean amplification up to drive level but, with plenty of gain early, blows away injection of power supply artefacts thus desensitising the whole amp to supply quality.
Lots of nice bouncy supply for the output stage so you take advantage of it's transient excess ability, then regulate the early stages to purify the lot. Not only does it make for nice clean amplification up to drive level but, with plenty of gain early, blows away injection of power supply artefacts thus desensitising the whole amp to supply quality.
amplifierguru said:
What is the point of having a 'nice bouncy supply for the output stage', if your gain stage cannot 'bounce' along with it to take advantage of the available 'bounce'?
In other words your unity-gain output will potentially 'bounce' to beyond the nominal rails, while your gain stage merely clips!!
I'll stick to unregulated through out.
Correction...
One detail you were right about RG:
Assuming the load demands 50A from the amp., then power P(av) dissipated in the MOSFET switch is given by:
P(av)~{(50/pi)^2}*Rds(on)
Assuming a typical Rds(on)~0.03Ohms, then:
P(av)<8W
Therefore, during normal use the MOSFET switch will require minimal heatsinking if at all....
One detail you were right about RG:
Assuming the load demands 50A from the amp., then power P(av) dissipated in the MOSFET switch is given by:
P(av)~{(50/pi)^2}*Rds(on)
Assuming a typical Rds(on)~0.03Ohms, then:
P(av)<8W
Therefore, during normal use the MOSFET switch will require minimal heatsinking if at all....
Hi Mikeks,
I did say "regulate the early stages" leaving the full swing driver stage AND output stage unregulated.
"I'll stick to unregulated throughout" That's OK as long as you match every stage's components, raise collector Z using cascoding and take diffl output and lose the miller C degradation -to hopefully eliminate supply artefacts.
Then there's still the induction. A couple of good low noise regulators is simpler AND more effective practically.
I did say "regulate the early stages" leaving the full swing driver stage AND output stage unregulated.
"I'll stick to unregulated throughout" That's OK as long as you match every stage's components, raise collector Z using cascoding and take diffl output and lose the miller C degradation -to hopefully eliminate supply artefacts.
Then there's still the induction. A couple of good low noise regulators is simpler AND more effective practically.
Mikeks,
Just have a look at the effect the miller C has on the vast majority of chip op amps - the PSRR taks a dive and in comes the supply hash. They use it to save chip space - it's one place a small C provides heaps of stability. What's your excuse?
My driver/outputs usually have gain, either in the output stage (inverted) or driver (complementary push pull). Then earlier stages can be a chip OR a discrete with limited swing.
Good regulator noise is not a problem in power amps but supply borne output commutation is! Using SE input stage output with simple RC filtering won't achieve the rejection of even a simple 2 transistor regulator.
Just have a look at the effect the miller C has on the vast majority of chip op amps - the PSRR taks a dive and in comes the supply hash. They use it to save chip space - it's one place a small C provides heaps of stability. What's your excuse?
My driver/outputs usually have gain, either in the output stage (inverted) or driver (complementary push pull). Then earlier stages can be a chip OR a discrete with limited swing.
Good regulator noise is not a problem in power amps but supply borne output commutation is! Using SE input stage output with simple RC filtering won't achieve the rejection of even a simple 2 transistor regulator.
amplifierguru said:
Actually I like the idea of regulating the low-current stages.
Also, the circuit below gives a substantial PSRR improvement. You may have to zoom in a bit to see the schematic without some lines looking like they are missing. See pages 248 and 249 of Self's book for more info. I think he got this idea from "A General Relationship Between Amplifier Parameters, And Its Application to PSRR Improvement" by Sackinger, Goette and Guggenbuhl, IEEE Transations on Circuits and Systems Vol 38, #10, Oct 1991 (thanks jcx for originally pointing out this paper to me). If you shoot me an email I can send it to you.
Attachments
Here's a few more , maybe of interest -
Power Supply Rejection in operational Transconductance Amplifiers by Steyaert & Sansen IEEE transactions on circuits and systems V37No9 Sept '90
On the relationship between the CMRR or PSRR and the second harmonic distortion of Differential Input amplifiers by Op'T EYNDE , WAMBACQ and SANSEN IEEE journal of solid state circuits V24No6 DEC '89
Sorry I only have poorly copied hard copy and no scanner. Good bedtime reading - asleep in seconds. Plenty more.
Power Supply Rejection in operational Transconductance Amplifiers by Steyaert & Sansen IEEE transactions on circuits and systems V37No9 Sept '90
On the relationship between the CMRR or PSRR and the second harmonic distortion of Differential Input amplifiers by Op'T EYNDE , WAMBACQ and SANSEN IEEE journal of solid state circuits V24No6 DEC '89
Sorry I only have poorly copied hard copy and no scanner. Good bedtime reading - asleep in seconds. Plenty more.
amplifierguru said:
The op amp in such an amplifier can have near ideal psrr (>150dB) across the audio band by using a simple RC filter...
The loss in voltage swing is not significant as this is compensated for by the gain in the output stage...
http://www.diyaudio.com/forums/showthread.php?s=&threadid=14482&highlight=
amplifierguru said:
I fail to see how supply rail rubbish can audibly affect an amp. with say -90dB PSRR across the audio band...This represents an attenuation greater than 30*10^-6!!
anatech said:
Stable quiescent conditions in your gain block are best maintained by constant current sources....
Mikeks,
My reference to Miller C comp and chips was that they were a fine example of the PSRR degradation of this compensation method.
My point was that this trivial type of compensation seriously degrades PSRR in typical se output to Vas amplifiers and only limited RC filtering is typically applied i.e. the common 100 ohm/100uF and diode. Just how much filtering does that provide at power bounce envelope frequencies of a soft supply?
RC can provide very good PSRR for chip driven amps >> 150dB agreed. Yes 90dB at least is necessary. NO BOUNCE.
My reference to Miller C comp and chips was that they were a fine example of the PSRR degradation of this compensation method.
My point was that this trivial type of compensation seriously degrades PSRR in typical se output to Vas amplifiers and only limited RC filtering is typically applied i.e. the common 100 ohm/100uF and diode. Just how much filtering does that provide at power bounce envelope frequencies of a soft supply?
RC can provide very good PSRR for chip driven amps >> 150dB agreed. Yes 90dB at least is necessary. NO BOUNCE.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.