Hi,
Does anyone have any views on increasing the bias in my Restek Fantasy amplifier. I've asked the question on the thread I started replating to the mods I've applied to the amp but I think the original title means I'm not getting the potential exposure i'd hoped for. I hope no one minds, but I'll repeat the question here.
The standard setting of 10mV seems a little on the low side to me after reading the TNT article on bias adjustment???
For reference, the schematic is here
Restek Fantasy Schematic
I've taken my measurement accross each of the output resistors which are 0.15r. (R5L1 to R5L4)
Based on this, I've started to increment the bias.
This table is based on 0.15r
1.5mV = 10mA
4 mV = 26.7mA
8 mV = 53.3mA
10mV = 66.7mA
15mV = 100mA
I got as far as 10mV but felt that the amp was running too hot under extended heavy load so I've backed it down to 7.5mV 50mA.
Is there any adverse effect to increasing the the bias other than heat which will ultimately lead to component failure in the output stage? I have to say that with the setting at 66mA, it was like listening to a completely different amp!!!!
Incidentally, one channel is constant accross all 4 resistors, and one channel varies but up to about 1.5mV. I'm guessing thats where the FETs in one channel are not well matched. How much of a difference will this make replacing with a matched set?
Thanks, Ian
Does anyone have any views on increasing the bias in my Restek Fantasy amplifier. I've asked the question on the thread I started replating to the mods I've applied to the amp but I think the original title means I'm not getting the potential exposure i'd hoped for. I hope no one minds, but I'll repeat the question here.
The standard setting of 10mV seems a little on the low side to me after reading the TNT article on bias adjustment???
For reference, the schematic is here
Restek Fantasy Schematic
I've taken my measurement accross each of the output resistors which are 0.15r. (R5L1 to R5L4)
Based on this, I've started to increment the bias.
This table is based on 0.15r
1.5mV = 10mA
4 mV = 26.7mA
8 mV = 53.3mA
10mV = 66.7mA
15mV = 100mA
I got as far as 10mV but felt that the amp was running too hot under extended heavy load so I've backed it down to 7.5mV 50mA.
Is there any adverse effect to increasing the the bias other than heat which will ultimately lead to component failure in the output stage? I have to say that with the setting at 66mA, it was like listening to a completely different amp!!!!
Incidentally, one channel is constant accross all 4 resistors, and one channel varies but up to about 1.5mV. I'm guessing thats where the FETs in one channel are not well matched. How much of a difference will this make replacing with a matched set?
Thanks, Ian
An externally hosted image should be here but it was not working when we last tested it.
If you call bias adjustment to preset wired to the base of the Vgamma multiplier at the base of the power stage, it only sets the quiescent current in them, the output offset voltage must be set in the first stage, normally a preset between the emitters of a differential pair. Keep in mind that a lower bias current will cause distortion at zero crossing signals, and too much will cause the output transistor to overheating because they go nearer class A.
Good luck!!!
Good luck!!!
Hi, yes that is exactly what I mean. So the only thing to be aware of is heat dissipation then (and obviously staying on the good side of an output stage disaster)?
Too much quiescent current can cause 'gm-doubling' distortion. Essentially, you move the crossover region away from zero crossing to higher voltage levels. This is theoretically wrong, but many people seem to prefer it - perhaps because the resultant distortion does not occur for very small signals.
Eh? Sorry, I'm not following.
I've not adjusted the DC offset. Only the Biasing current to the base of the output transistors. The DC offset has always been virtually nothing. I think there may be a servo becuase I cant see another adjustment and no matter what, DC on the output is virtually nothing.
I've not adjusted the DC offset. Only the Biasing current to the base of the output transistors. The DC offset has always been virtually nothing. I think there may be a servo becuase I cant see another adjustment and no matter what, DC on the output is virtually nothing.
In push-pull Class B most of the output waveform is handled by just one half of the output stage, the other being cut off. Near the zero crossing of the signal both halves of the output stage are working, so you get gain from both. In an ideal world you could arrange things so that as the signal voltage sweeps towards and then past zero the gain curves for the output stage would exactly match. For example, at the zero crossing you want each half to contribute half the usual gain, so the total is the same as when only one half is working. You can get close to this ideal but you can't get there exactly - the two curves don't match so in the central region you get gain wiggles. Global NFB then hides them. The snag is that you could get some distortion in this low-signal region, especially if the bias drifts a little due to temperature fluctuations. You could get either too much or too little gain around the zero crossing.
To avoid this, some people advocate a significant increase in bias current. This guarantees that in and around the central region you get twice the transconductance gain from the output stage. This 'gm-doubling' region is made wide enough to cover most low level signals, so there is not distortion. The snag is that for bigger signals there still has to be a transition from both halves working to one half working, otherwise its not Class B. This transition just happens partway up the graph instead of at the origin. You get a change in gain, but again the NFB hides it.
I can't comment on the sound. I would expect to see a small increase in distortion for mid-level signals, but a reduction in distortion for small signals. This trick can only be done if the amp has plenty of negative feedback to hide the transitions. In an ideal world, correct bias would eliminate crossover distortion whether NFB is present or not. Deliberately wrong bias will necessarily create distortion even in an ideal world.
To avoid this, some people advocate a significant increase in bias current. This guarantees that in and around the central region you get twice the transconductance gain from the output stage. This 'gm-doubling' region is made wide enough to cover most low level signals, so there is not distortion. The snag is that for bigger signals there still has to be a transition from both halves working to one half working, otherwise its not Class B. This transition just happens partway up the graph instead of at the origin. You get a change in gain, but again the NFB hides it.
I can't comment on the sound. I would expect to see a small increase in distortion for mid-level signals, but a reduction in distortion for small signals. This trick can only be done if the amp has plenty of negative feedback to hide the transitions. In an ideal world, correct bias would eliminate crossover distortion whether NFB is present or not. Deliberately wrong bias will necessarily create distortion even in an ideal world.
Thanks, an excellent explanation
OK, so increasing the bias will move the changeover of 2 halves working to 1 half working is further away from the crossover point.
So my experience in increasing the bias was a dramatic increase audio performance especially at lower levels and the fact that instrument placement became more precise would match what you are saying.
Why would the manufacturer have set the bias so low from the factory? Is there a way to measure/set the ideal point or is it down to individual users perception?
Also there seems to be some confusion over the use of the term DC offset. Is this purely the amount of DC present on the output (ideally 0) or as suggested, does it refer to differences between the output transistors meaning 1 could be working harder then the other (assuming this would be caused by unmatched transistor pairs?)
Thanks
OK, so increasing the bias will move the changeover of 2 halves working to 1 half working is further away from the crossover point.
So my experience in increasing the bias was a dramatic increase audio performance especially at lower levels and the fact that instrument placement became more precise would match what you are saying.
Why would the manufacturer have set the bias so low from the factory? Is there a way to measure/set the ideal point or is it down to individual users perception?
Also there seems to be some confusion over the use of the term DC offset. Is this purely the amount of DC present on the output (ideally 0) or as suggested, does it refer to differences between the output transistors meaning 1 could be working harder then the other (assuming this would be caused by unmatched transistor pairs?)
Thanks
DC offset is the DC voltage at the output terminals. It should be as near zero as possible. The main source of offset is unbalance in the input pair.
Setting output bias exactly is difficult. In theory a calculation or simulation can give a figure, which depends greatly on the output stage topology (low current for complementary pair, high current for Darlington pair). Even this 'ideal' current depends on assumptions about what sort of distortion you prefer, as zero distortion is not an option. Temperature makes a big difference to the current, and a small difference to the required current. Fortunately NFB means that the actual setting is not as critical as it otherwise would be.
See Doug Self's book on power amps for more on this, even if you disagree with his conclusions.
Setting output bias exactly is difficult. In theory a calculation or simulation can give a figure, which depends greatly on the output stage topology (low current for complementary pair, high current for Darlington pair). Even this 'ideal' current depends on assumptions about what sort of distortion you prefer, as zero distortion is not an option. Temperature makes a big difference to the current, and a small difference to the required current. Fortunately NFB means that the actual setting is not as critical as it otherwise would be.
See Doug Self's book on power amps for more on this, even if you disagree with his conclusions.
Sounds to me like so long as you are sensible about the adjustments and keep an eye on the temperature of the output devices, its down to your ear then!
I've found Doug Self's website too (along with his book on Amazon) so I'll be taking a read.
Thanks for the input, very much appreciated...
Ian
I've found Doug Self's website too (along with his book on Amazon) so I'll be taking a read.
Thanks for the input, very much appreciated...
Ian
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.