Ian,
I've sent the jpg diagram via webmaster. My (old) system runs into problems when I try to downsize file size.
I know the Plitron o/p tranny on the 4040 has an exceptional LF bandwidth. I wonder what the leakage inductance figures are The fact that some users reported HF oscillations is sure sign that amp is load touchy. I notice C15 (22nF) as intercoupler on concertina. This suprises me as the voltages either side are nearly the same. On phaseshift grounds with other designs this cap was eliminated so I wonder.
richj
I've sent the jpg diagram via webmaster. My (old) system runs into problems when I try to downsize file size.
I know the Plitron o/p tranny on the 4040 has an exceptional LF bandwidth. I wonder what the leakage inductance figures are The fact that some users reported HF oscillations is sure sign that amp is load touchy. I notice C15 (22nF) as intercoupler on concertina. This suprises me as the voltages either side are nearly the same. On phaseshift grounds with other designs this cap was eliminated so I wonder.
richj
richj,
Counter intuitively (or perhaps not), the better the output transformer the more likely you are to have parasitic oscillation problems in the output stage - especially in Ultralinear Mode.
Changing those 180R screen resistors to 1K is a MUST. Also check the PCB layout, if those resitors are NOT right against the EL34 screen pin then replace the resistors with wire links, cut tracks to the screen pin and add new screen resistors across the track cut.
A rough calc shows that those 22nF coupling caps will give a LF roll off (-3dB) at about 20Hz which seems very high. This may be deliberate - to stabilize the amp at low freqencies (i.e. to get away with less/cheaper power supply decoupling).
The C15 22nF will give a 9Hz LF corner with R23 - probably why the 22nF are used for the coupling caps, to keep the corner frequencies (zeros) separted although X2 is really NOT ENOUGH.
I'd be tempted to ditch R23, replace C15 with a short circuit (that is DC couple the stages) and than adjust the input stage current so that the voltage into the Concertina grid is correct.
I looked at buying one of these a few years ago but decided against it when I looked at the schematic, just too many "cut corners". Having said that I think they would make an excellent candidate for rebuilding into something very good indeed.
Cheers,
Ian
Counter intuitively (or perhaps not), the better the output transformer the more likely you are to have parasitic oscillation problems in the output stage - especially in Ultralinear Mode.
Changing those 180R screen resistors to 1K is a MUST. Also check the PCB layout, if those resitors are NOT right against the EL34 screen pin then replace the resistors with wire links, cut tracks to the screen pin and add new screen resistors across the track cut.
A rough calc shows that those 22nF coupling caps will give a LF roll off (-3dB) at about 20Hz which seems very high. This may be deliberate - to stabilize the amp at low freqencies (i.e. to get away with less/cheaper power supply decoupling).
The C15 22nF will give a 9Hz LF corner with R23 - probably why the 22nF are used for the coupling caps, to keep the corner frequencies (zeros) separted although X2 is really NOT ENOUGH.
I'd be tempted to ditch R23, replace C15 with a short circuit (that is DC couple the stages) and than adjust the input stage current so that the voltage into the Concertina grid is correct.
I looked at buying one of these a few years ago but decided against it when I looked at the schematic, just too many "cut corners". Having said that I think they would make an excellent candidate for rebuilding into something very good indeed.
Cheers,
Ian
Re: Concertina Bias
It’s a possibility that a small grid current in ECC83 will make a voltage drop over R23, but it can’t be many mV bias, can it?
The schematic tells us a 100V drop over the 47k cathode resistor, that’s ~2mA and the cathode/anode voltage over ECC83 is ~180V. These values should indicate a bias more like ~1V. It’s not clear to me that the 820k grid leak can supply that.
gingertube said:
It’s a possibility that a small grid current in ECC83 will make a voltage drop over R23, but it can’t be many mV bias, can it?
The schematic tells us a 100V drop over the 47k cathode resistor, that’s ~2mA and the cathode/anode voltage over ECC83 is ~180V. These values should indicate a bias more like ~1V. It’s not clear to me that the 820k grid leak can supply that.
gingertube said:
Well, not really.
Since R23 is connected to the cathode, the fc depends on the amplification in the split load stage. If the amplification has been 1, the apparent value of R23 would have been infinite. The amplification factor is more like 0.95, which means that the apparent value of R23 is many times higher than 820k and the fc lower than 9Hz.
Pouring cold water on modification ideas ???.this thread has gotton quiet...The 4040 users club is going to need help alright. Someone out there must have a chassis that works at full power „or sort of“ ??
Now the bad news. (only an idea) If I had one and was stuck with it destroying output tubes, I would be tempted to try 6L6's or better still 5881's which have lower gm than the 34's or 88 class. The 6L6 family will require a little more drive voltage but o/p stage will be more tamer with less o/p power. It will require readjustment of neg grid bias if enough range exists. Another problem with the 4040 is that high value grid leaks are used on the output stage (g1) to avoid increased thd from stage loading the concertina.
Having serious doubts with the existing design configuration, short of complete butchery and starting again with a fresh design; ...the o/p tube change is a possible way out but isn’t without snags. It may not resolve the problems but other members could comment here.. Fixed bias does require bias resistor values to be way lower than auto bias.
I wonder if anyone has successfully/tried converting the 4040 to autobias.
With the right value cathode decoupling cap this would automatically add a 6dB rolloff pole in the o/p stage circuit and output stage grid input resistor values would be less critical and remain as they are. From that aspect autobias is attactive, has lower output impedance and more longterm stable but with lower output power.
The other point of any amp design is the value of the o/p stage grid leak resistors. With modern-day tubes outgassing isn't as deep as it was per early days and this requires lower values of grid leak resistor. Even Svetlana KT88 types claim around 100K for fixed bias and I wouldn't run an EL34 anything higher.The onus is then on the driver circuit capable of supplying the goods into a lowish impedance load with low thd. With low impedance drivers running into parallel paired output stage grids, Miller effect will be well controlled. That means for a p-p 150W amp using 4x KT88‘s, a balanced power voltage driver capable of giving 40+40V RMS into a 50Kohm load with B+ of 360V, thd will be around 1% mark. Forget cathode followers. The design of the driver stage is important for reliably driving output stage tubes and an ECC83 isn’t the right tube.
The polished chrome on a 4040 doesn't marry with the under- performed and under-optimal circuit. Signal to noise may be good but the rest of the details espec transient response is questionable. I certainly don’t agree with tapped secondary windings for a parallel push pull under difficult load conditions. Short of a complete rebuild it‘s back to the beginning.
richj
Now the bad news. (only an idea) If I had one and was stuck with it destroying output tubes, I would be tempted to try 6L6's or better still 5881's which have lower gm than the 34's or 88 class. The 6L6 family will require a little more drive voltage but o/p stage will be more tamer with less o/p power. It will require readjustment of neg grid bias if enough range exists. Another problem with the 4040 is that high value grid leaks are used on the output stage (g1) to avoid increased thd from stage loading the concertina.
Having serious doubts with the existing design configuration, short of complete butchery and starting again with a fresh design; ...the o/p tube change is a possible way out but isn’t without snags. It may not resolve the problems but other members could comment here.. Fixed bias does require bias resistor values to be way lower than auto bias.
I wonder if anyone has successfully/tried converting the 4040 to autobias.
With the right value cathode decoupling cap this would automatically add a 6dB rolloff pole in the o/p stage circuit and output stage grid input resistor values would be less critical and remain as they are. From that aspect autobias is attactive, has lower output impedance and more longterm stable but with lower output power.
The other point of any amp design is the value of the o/p stage grid leak resistors. With modern-day tubes outgassing isn't as deep as it was per early days and this requires lower values of grid leak resistor. Even Svetlana KT88 types claim around 100K for fixed bias and I wouldn't run an EL34 anything higher.The onus is then on the driver circuit capable of supplying the goods into a lowish impedance load with low thd. With low impedance drivers running into parallel paired output stage grids, Miller effect will be well controlled. That means for a p-p 150W amp using 4x KT88‘s, a balanced power voltage driver capable of giving 40+40V RMS into a 50Kohm load with B+ of 360V, thd will be around 1% mark. Forget cathode followers. The design of the driver stage is important for reliably driving output stage tubes and an ECC83 isn’t the right tube.
The polished chrome on a 4040 doesn't marry with the under- performed and under-optimal circuit. Signal to noise may be good but the rest of the details espec transient response is questionable. I certainly don’t agree with tapped secondary windings for a parallel push pull under difficult load conditions. Short of a complete rebuild it‘s back to the beginning.
richj
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.