I've just built this amp - schematic enclosed. Problem I'm having is distortion on both channels. voltages seem correct - except output 6b4gs are not perfectly matched - filaments range from 5 to 6.5 volts. Bought new in russia!
Power supply is on a separate chassis, and is nice enough - cap>choke>cap with last cap being polypropylene. 80uF cathode bypass on 6B4Gs also polypropylene. One PSU for outputs, another for input and driver.
What can be causing this high distortion - sounds horrible!
thanks for any suggestions on the circuit.
andy evans
Power supply is on a separate chassis, and is nice enough - cap>choke>cap with last cap being polypropylene. 80uF cathode bypass on 6B4Gs also polypropylene. One PSU for outputs, another for input and driver.
What can be causing this high distortion - sounds horrible!
thanks for any suggestions on the circuit.
andy evans
Overall distortion could be due to (the order is not relevant):
1) Excessive distortion from the driver. It could be unbalanced and/or 27K a bit low for the max swing you need
2) Transformer primary load (with auto-bias I would suggest higher load than typical 5K)
3) Substantial unbalance of the output stage (it can also be just a consequence of different filament operative conditions)
5V rather than 6.3V for the filament voltage will alter considerably the sound.
Sometimes a lower voltage (say 5.7V instead of 6.3V) results in a better sound. Such a difference is more than one could expect....
45
1) Excessive distortion from the driver. It could be unbalanced and/or 27K a bit low for the max swing you need
2) Transformer primary load (with auto-bias I would suggest higher load than typical 5K)
3) Substantial unbalance of the output stage (it can also be just a consequence of different filament operative conditions)
5V rather than 6.3V for the filament voltage will alter considerably the sound.
Sometimes a lower voltage (say 5.7V instead of 6.3V) results in a better sound. Such a difference is more than one could expect....
45
Q and As
I can increase the driver anode resistor from 27k to 38k but at a slightly lower current - does this sound preferable? would be same as first stage.
Do I need to decouple first and second stages?
OPt is in fact 6.6K
For the 6B4G filaments I have a fixed sense resistor on the LM1084 current source. I could add a potentiomenter in parallel with that to give me some adjustment
andy
I can increase the driver anode resistor from 27k to 38k but at a slightly lower current - does this sound preferable? would be same as first stage.
Do I need to decouple first and second stages?
OPt is in fact 6.6K
For the 6B4G filaments I have a fixed sense resistor on the LM1084 current source. I could add a potentiomenter in parallel with that to give me some adjustment
andy
Distortion is on both channels but worse on one. If I put on a choral work (Bach Cantata) the sound breaks up. The chorus sounds a bit like broken glass. Distortion is continuous. I'm playing through a single 6" driver since my Apogee ribbons aren't sensitive enough.
I have a scope but don't know how to use it yet!
andy
I have a scope but don't know how to use it yet!
andy
andyjevans said:
38K with slightly lower current would be better, IMHO.
Decoupling the second stage will bring an improvement.
You may also want to use a direct coupling between 1st and 2nd stage but you need very high supply voltage, otherwise you will not find a good operative solution for the 2nd stage.
I think it could be a very good idea.
In my experience many 6SN7's are not so good at low levels. Basically the output is affected by a constant 0.5%-0.7% THD from 1 Vrms up to several volts and then distortion starts rising as one would expect. Using a cascade made of two identical (directly coupled) tubes you end up with quite an improvement.
You may want to add a small value (1 ohm - 1% tolerance) between each plate and its transformer connection to check the anode current precisely. It will not affect anything else.
Have you tried to remove the LM1084? If not, try it possibly adding a small resistor in series with filament. This will help a little bit to limit the current when the filament is cold. don't worry if you get 6V. It is important you stay within the specification (6.3V +/- 10%).
45
Even if the 6SN7 is not optimum, if everything is biased properly you shouldn't have constant distortion. Have you checked the operating points of each stage? Maybe there is a miscalculation and you are clipping a tube input somewhere? My guess would be at the second stage.
Maybe your shunt volume arrangement isn't working as you expect?
Is the first 6SN7 splitting phase? Or are you feeding it balanced? I can't really tell from the schematic. A trick to making the attachment smaller while preserving its display size is to save it as a black and white TIFF file instead of a jpg.
Maybe your shunt volume arrangement isn't working as you expect?
Is the first 6SN7 splitting phase? Or are you feeding it balanced? I can't really tell from the schematic. A trick to making the attachment smaller while preserving its display size is to save it as a black and white TIFF file instead of a jpg.
Using a ccs to heat filaments in parallel is a good way to assure that the currents actually don't share properly.
* Filaments wired in parallel should be heated with constant voltage.
* Filaments wired in series should be heated with constant current.
* Individual filaments may be heated either way as technical requirements or preferences dictate.
Measure all voltages, and the voltage across the output stage cathode bias resistor. Learn to model your circuit in LTSpice IV, (free) note that 6B4G can be modeled as a 2A3.
I have used a similar driver topology based on 6SL7 and 6SN7 in cascade dc coupled to drive 300B in pushpull to 30Wrms. Refer to my ancient VTV article which IIRC is in issue 11 published in 1998 or thereabouts. (I could have the wrong issue # ) The design is also available on my site, but is non-free at this point.
The tail resistances in your design are so small that I doubt these stages function well as differential amplifiers, their common mode rejection will be quite poor. Much larger tail resistors are required and these would usually need to returned to a negative supply. (Alternatives exist, but in the first stage would require input coupling caps, and one extra resistor for the bias in the cathode circuit is required..) Seems like this is a spot where a CCS could be used to advantage, however a small negative supply is still probably required.
See one option below... (Would work well in the second stage with minor changes.)
* Filaments wired in parallel should be heated with constant voltage.
* Filaments wired in series should be heated with constant current.
* Individual filaments may be heated either way as technical requirements or preferences dictate.
Measure all voltages, and the voltage across the output stage cathode bias resistor. Learn to model your circuit in LTSpice IV, (free) note that 6B4G can be modeled as a 2A3.
I have used a similar driver topology based on 6SL7 and 6SN7 in cascade dc coupled to drive 300B in pushpull to 30Wrms. Refer to my ancient VTV article which IIRC is in issue 11 published in 1998 or thereabouts. (I could have the wrong issue # ) The design is also available on my site, but is non-free at this point.
The tail resistances in your design are so small that I doubt these stages function well as differential amplifiers, their common mode rejection will be quite poor. Much larger tail resistors are required and these would usually need to returned to a negative supply. (Alternatives exist, but in the first stage would require input coupling caps, and one extra resistor for the bias in the cathode circuit is required..) Seems like this is a spot where a CCS could be used to advantage, however a small negative supply is still probably required.
See one option below... (Would work well in the second stage with minor changes.)
Attachments
Wow - some nice input - thanks guys.
First - the 6B4G filaments are not in parallel - I only use individual filament supplies these days. But they clearly need some trimmer to bring the voltages right.
Second the shunt volume control. i was trying to be clever and using the in line resistor before the shunt volume control as a grid stopper. I've done this in the past and it works, but as has been pointed out to me, there's the possibility of RF on the wires going to the shunt volume control which are on the grid pins. So this looks like not so clever - back to the textbook
Third, the output is unnecessarily wimpy given that Sovtek 6B4Gs can take something like I guess 350v, so I should ramp up the HT for the final stage.
Fourth - looks like I should use CCSs under the diff pairs. I usually do but was put off by Lynn Olsen saying he didn't like them. Can't remember that post very well - maybe you can!
Will get the soldering iron out and report on changes.
andy
First - the 6B4G filaments are not in parallel - I only use individual filament supplies these days. But they clearly need some trimmer to bring the voltages right.
Second the shunt volume control. i was trying to be clever and using the in line resistor before the shunt volume control as a grid stopper. I've done this in the past and it works, but as has been pointed out to me, there's the possibility of RF on the wires going to the shunt volume control which are on the grid pins. So this looks like not so clever - back to the textbook
Third, the output is unnecessarily wimpy given that Sovtek 6B4Gs can take something like I guess 350v, so I should ramp up the HT for the final stage.
Fourth - looks like I should use CCSs under the diff pairs. I usually do but was put off by Lynn Olsen saying he didn't like them. Can't remember that post very well - maybe you can!
Will get the soldering iron out and report on changes.
andy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.