I would appreciate some comments on this amplifier from anybody with experience, please.
As has been found previously the schematic was impossible to find so I plotted one out myself (a model presently in my possession was fortunately in top shape). It is different from the neighbouring models MQ50 and MQ80 I could find. The input stage ECC83 is an ltp with global feedback going to the bottom triode, and with a relatively large tail of several 100K (fed from a negative supply). This is directly coupled to the ECC85 driver with common cathode resistor (12K), in its turn C-feeding p.p. output EL34s with both some cathode feedback and UL taps.
A practical problem I found is the relatively high gain input diff. triode, directly coupled to a succeeding stage, also of moderately high gain. Admittedly the d.c. coupling takes the form of an l.f. step in the form of a 1M resistor bypassed by a 470nF capacitor each side, ECC85 grids 'tied together' by another 1M resistor. That decreases the d.c. gain, but I still have difficulty in keeping the ECC85 triodes balanced over time. A g1-input difference of >1,5V starts steering their plates towards non-linear regions - some >25V plate voltage difference - and maintaining ECC83 anodes to within that potential equality particularly over time, I'm afraid does not seem to work out all that well in practice. (All anode loads have balance pots in their circuits, still.)
One realises the purpose of having a gain/phase reduction step in the l.f. mix, but I fear a further r.c. coupling might have to be brought in, hopefully formatable so as not to compromise l.f. stability. Fortunately components are smaller these days.
Anybody with wisdom/experience on this? (Why do I get the impression that this Luxman model was not popular?)
Thanks!
As has been found previously the schematic was impossible to find so I plotted one out myself (a model presently in my possession was fortunately in top shape). It is different from the neighbouring models MQ50 and MQ80 I could find. The input stage ECC83 is an ltp with global feedback going to the bottom triode, and with a relatively large tail of several 100K (fed from a negative supply). This is directly coupled to the ECC85 driver with common cathode resistor (12K), in its turn C-feeding p.p. output EL34s with both some cathode feedback and UL taps.
A practical problem I found is the relatively high gain input diff. triode, directly coupled to a succeeding stage, also of moderately high gain. Admittedly the d.c. coupling takes the form of an l.f. step in the form of a 1M resistor bypassed by a 470nF capacitor each side, ECC85 grids 'tied together' by another 1M resistor. That decreases the d.c. gain, but I still have difficulty in keeping the ECC85 triodes balanced over time. A g1-input difference of >1,5V starts steering their plates towards non-linear regions - some >25V plate voltage difference - and maintaining ECC83 anodes to within that potential equality particularly over time, I'm afraid does not seem to work out all that well in practice. (All anode loads have balance pots in their circuits, still.)
One realises the purpose of having a gain/phase reduction step in the l.f. mix, but I fear a further r.c. coupling might have to be brought in, hopefully formatable so as not to compromise l.f. stability. Fortunately components are smaller these days.
Anybody with wisdom/experience on this? (Why do I get the impression that this Luxman model was not popular?)
Thanks!
Last edited:
Hello,
I happen to have an MQ70 on my desk for repair. The symptom is random noises coming from both channels. Already replaced the ECC83 plate resistors: less noise now, but still some. And I can't balance the plate voltages of the ECC83, so it has to be replaced.
Peculiar circuit, but I must say I don't like it, with all those high-value resistors getting hot and looking for trouble. Looks like a BJT amplifier made with tubes, with the differential first stage as an error amplifier, second stage as a VAS, and output stage.
I confirm your impression of poor sales success of this amp: it's the first time I've ever seen one.
I happen to have an MQ70 on my desk for repair. The symptom is random noises coming from both channels. Already replaced the ECC83 plate resistors: less noise now, but still some. And I can't balance the plate voltages of the ECC83, so it has to be replaced.
Peculiar circuit, but I must say I don't like it, with all those high-value resistors getting hot and looking for trouble. Looks like a BJT amplifier made with tubes, with the differential first stage as an error amplifier, second stage as a VAS, and output stage.
I confirm your impression of poor sales success of this amp: it's the first time I've ever seen one.
Last edited:
Johan,
What are the O/P tubes? What do the PSUs look like? Is the "iron" decent? Can you upload a photo or 2? Maybe we can rationalize the circuitry of the "beast".
You mentioned LTP and negative PSU. Replacing the tail resistor with a CCS will force symmetry between the 2 sides. FWIW, we demonstrated in "El Cheapo" that a good CCS in a LTP's tail makes applying NFB to the non-inverting triode's grid work WELL.
What are the O/P tubes? What do the PSUs look like? Is the "iron" decent? Can you upload a photo or 2? Maybe we can rationalize the circuitry of the "beast".
You mentioned LTP and negative PSU. Replacing the tail resistor with a CCS will force symmetry between the 2 sides. FWIW, we demonstrated in "El Cheapo" that a good CCS in a LTP's tail makes applying NFB to the non-inverting triode's grid work WELL.
If Johan has a digital camera, he knows my email address and I can upload his pics.
bulgin
The tubes line up are: 12AX7, 6AQ8 and 6CA7/EL34. The OPT has UL tap and CF windings. One Japanese page mentioned changing the tail resistor for the 12AX7 differential amplifier to a CRD/CCS, also replacing the coupling network between the 12AX7 and 6AQ8 with just resistors (grid stoppers). Hope this is of some use.
With both UL taps and CF windings, I can see where LOTS of small signal gain would be needed.
The 6AQ8/ECC85 strongly resembles the 12AT7/ECC81.
While the "El Cheapo" LTP splitter/driver comes up short in the gain dept. for driving the EL34 class "finals", adding an additional voltage gain stage outside the NFB loop disposes of the issue. A CCS loaded triode section from the 12AY7, 5751, 12AX7 group, depending on how sensitive the amp needs to be, cap. coupled to the LTP would handle that need. Wiring 1 channel to use system I and the 2nd channel to use system II doubles tube life, via cross swapping, and saves 300 mA. of heater current. Another linear, heater current saving, option for the no loop NFB I/P stage is an EF86 with an unbypassed cathode bias resistor. There's much to be said in favor of pentode I/P stages, in power amps.
Getting the low gm type out of the NFB loop would improve slew limiting behavior. High pass filtration at the amp's I/Ps protects against O/P "iron" core saturation.
The 6AQ8/ECC85 strongly resembles the 12AT7/ECC81.
While the "El Cheapo" LTP splitter/driver comes up short in the gain dept. for driving the EL34 class "finals", adding an additional voltage gain stage outside the NFB loop disposes of the issue. A CCS loaded triode section from the 12AY7, 5751, 12AX7 group, depending on how sensitive the amp needs to be, cap. coupled to the LTP would handle that need. Wiring 1 channel to use system I and the 2nd channel to use system II doubles tube life, via cross swapping, and saves 300 mA. of heater current. Another linear, heater current saving, option for the no loop NFB I/P stage is an EF86 with an unbypassed cathode bias resistor. There's much to be said in favor of pentode I/P stages, in power amps.
Getting the low gm type out of the NFB loop would improve slew limiting behavior. High pass filtration at the amp's I/Ps protects against O/P "iron" core saturation.
Gentlemen,
Looked here only now. The tube line-up has been posted. The iron is quite decent. I have not been able to balance the pre-stages to satisfaction and "chickened out" by aborting the direct coupling between the input ECC83 to isolate the ECC85 driver to its own bias. (Omitted the 1M resistors in parallel with the 470nF caps - see later).
And therewith lies my "defeat". I almost PM'd SY, but knew the answer. As noticed in the first post, I simply could not get the balance forced by Kirchoff to work although I am conversant with the theory.
Firstly: The circuit uses the common CCS principle by way of a large common cathode resistor (220K) fed from a zener-stabilized -220V supply (figure from memory; amp no longer with me). I configured that this is close enough to a CCS for all practical purposes.
To lay my embarrassment before 'the nation' now: This is supposed to render equal ECC83 anode voltages. Direct-coupling to the ECC85 grids should then render equal anode voltages there, again courtesy of its CCS. But it does not nearly happen, even with an equalising pot in the --83 anode circuit. (Other matters taken care of: Equal resistors, good tubes etc.) My problem: I know balance should be forced by K-Law: on the other hand I do not get further than seeing that K-Law forces the sum of the anode currents to be constant. Repeat: I know SY has demonstrated this; what to blame for not seeing this? ... age, dementia ...
Perhaps a PM to Eli or Sy to help out.
To cut short: I abandoned the present schematic, ac coupled to the ECC85 by removing the 1M resistors. This stage could then work with normal cathode bias and the amplifier 'restored' to good operation. (It was a restortion and had to be finished.)
To sum up: I could not get the ECC83 anode +120V to within the <1V required to have the ECC85 within its operating zone, despite the balancing pots provided. I grounded the ECC85 grids through 1M, common cathode resistor 390 ohm, resultant anode voltages 274Vdc, retaining the anode resistors and pot provided and adjusting to equal output signal.
Extra note: I also replaced coupling caps with 630V types; previous 400V rating was temporarily exceeded during warm-up.
Orit,
Thus not helping you in a correct way; I can also not explain the noise other than coming from the three 67V(?) zeners in the ltp circuit. In my model there was no noise.
___________________________________________________________________________
Concur with Eli.
Saw the post only after I submitted mine. Yes, the power cathode feedback does require a higher G1 drive, thus the problem of getting the ECC85 anode voltages within a narrow margin to provide for that.
Looked here only now. The tube line-up has been posted. The iron is quite decent. I have not been able to balance the pre-stages to satisfaction and "chickened out" by aborting the direct coupling between the input ECC83 to isolate the ECC85 driver to its own bias. (Omitted the 1M resistors in parallel with the 470nF caps - see later).
And therewith lies my "defeat". I almost PM'd SY, but knew the answer. As noticed in the first post, I simply could not get the balance forced by Kirchoff to work although I am conversant with the theory.
Firstly: The circuit uses the common CCS principle by way of a large common cathode resistor (220K) fed from a zener-stabilized -220V supply (figure from memory; amp no longer with me). I configured that this is close enough to a CCS for all practical purposes.
To lay my embarrassment before 'the nation' now: This is supposed to render equal ECC83 anode voltages. Direct-coupling to the ECC85 grids should then render equal anode voltages there, again courtesy of its CCS. But it does not nearly happen, even with an equalising pot in the --83 anode circuit. (Other matters taken care of: Equal resistors, good tubes etc.) My problem: I know balance should be forced by K-Law: on the other hand I do not get further than seeing that K-Law forces the sum of the anode currents to be constant. Repeat: I know SY has demonstrated this; what to blame for not seeing this? ... age, dementia ...
Perhaps a PM to Eli or Sy to help out.To cut short: I abandoned the present schematic, ac coupled to the ECC85 by removing the 1M resistors. This stage could then work with normal cathode bias and the amplifier 'restored' to good operation. (It was a restortion and had to be finished.)
To sum up: I could not get the ECC83 anode +120V to within the <1V required to have the ECC85 within its operating zone, despite the balancing pots provided. I grounded the ECC85 grids through 1M, common cathode resistor 390 ohm, resultant anode voltages 274Vdc, retaining the anode resistors and pot provided and adjusting to equal output signal.
Extra note: I also replaced coupling caps with 630V types; previous 400V rating was temporarily exceeded during warm-up.
Orit,
Thus not helping you in a correct way; I can also not explain the noise other than coming from the three 67V(?) zeners in the ltp circuit. In my model there was no noise.
___________________________________________________________________________
Concur with Eli.
Saw the post only after I submitted mine. Yes, the power cathode feedback does require a higher G1 drive, thus the problem of getting the ECC85 anode voltages within a narrow margin to provide for that.
Last edited:
jazbo8 posted the schematic in another thread before but let's repost it here for easier reference:
An externally hosted image should be here but it was not working when we last tested it.
Thanks Directdriver!
Should have included that myself. To annotate a few values for extra clarity - these are what I measured on my model (pity no component designations are indicated, I will therefore describe):
Power Supply:
V(B1) = 480V, V(B2) = 460V, V(B3) = 210V.
Grid bias bottom of 47µF: -259V
The three diodes in parallel to the 150K resistors: Zener diodes 1N4760A, 68V. Diode feeding point C2 = T86; ordinary (Schottky?) diode.
Voltage at junction of 30K EL34 bias pots: -49V
ECC83 cathodes: +1,3V --- ECC83 anodes: +135V
ECC85 cathodes: +143V --- ECC85 anodes: + 378V*
*This was too high to enable feeding the EL34s; the wave top flattened out before max. output was reached. Also these voltages are averaged; as said I could not reach balance originally.
After my changes (Error before: The ECC85 new cathode resistor to common was 220 ohm, not 390 ohm as mentioned earlier.) and with 47K ECC85 load resistors;
ECC85 anodes = +220V, ECC85 cathodes = +2,2V
Under these conditions I found 40W output into 8 ohm load with some headroom at ECC85 outputs. As Jazbo mentioned, some extra voltage is required here to provide for EL34 UL operation plus some 8,5% cathode feedback.
(Note that the cathode fed-back signal is part of the (screen to B+) UL signal, thus screen taps here lower than the conventional 40%.)
Hope this gives further elucidation
- and I would still welcome other experiences. How was the original supposed to work? What am I not seeing?
Should have included that myself. To annotate a few values for extra clarity - these are what I measured on my model (pity no component designations are indicated, I will therefore describe):
Power Supply:
V(B1) = 480V, V(B2) = 460V, V(B3) = 210V.
Grid bias bottom of 47µF: -259V
The three diodes in parallel to the 150K resistors: Zener diodes 1N4760A, 68V. Diode feeding point C2 = T86; ordinary (Schottky?) diode.
Voltage at junction of 30K EL34 bias pots: -49V
ECC83 cathodes: +1,3V --- ECC83 anodes: +135V
ECC85 cathodes: +143V --- ECC85 anodes: + 378V*
*This was too high to enable feeding the EL34s; the wave top flattened out before max. output was reached. Also these voltages are averaged; as said I could not reach balance originally.
After my changes (Error before: The ECC85 new cathode resistor to common was 220 ohm, not 390 ohm as mentioned earlier.) and with 47K ECC85 load resistors;
ECC85 anodes = +220V, ECC85 cathodes = +2,2V
Under these conditions I found 40W output into 8 ohm load with some headroom at ECC85 outputs. As Jazbo mentioned, some extra voltage is required here to provide for EL34 UL operation plus some 8,5% cathode feedback.
(Note that the cathode fed-back signal is part of the (screen to B+) UL signal, thus screen taps here lower than the conventional 40%.)
Hope this gives further elucidation
- and I would still welcome other experiences. How was the original supposed to work? What am I not seeing?
Last edited:
Man, I had forgotten all about that thread, nor remember how I got all that information, getting senile...
Last edited:
Hi (excuse me I was not on DiyAudio these days). Thank you for the hint, the noise really seems to come from the zeners, specifically from the most negative one, as the noise disappers by applying some pressure on it.
Actually it's a kind of noise I associate with silicon devices. Off for some new zeners, before shops close for holidays ...
Gentlemen,
shortly after my previous post, I found the real culprit for the random noise: both ECC85/6AQ8 for L and R channels replaced, noise gone. They have probably been stressed by the abnormal Heater-Cathode voltage of this (ill) design, 150 volts against 90 max as the datasheets say. Obviously, I expect even these replaced ECC85's will soon start to be noisy. I wonder how did Luxman's QC department pass this project.
shortly after my previous post, I found the real culprit for the random noise: both ECC85/6AQ8 for L and R channels replaced, noise gone. They have probably been stressed by the abnormal Heater-Cathode voltage of this (ill) design, 150 volts against 90 max as the datasheets say. Obviously, I expect even these replaced ECC85's will soon start to be noisy. I wonder how did Luxman's QC department pass this project.
Orit,
Indeed. See my previous notes. As said, I found the anode-cathode voltage(s) for the ECC85 too low to 'fit in' the signal required for the EL34s without 'bottoming'.
Cannot recall now but ..... oh, of course. I changed the circuit to ac-coupled, so that the ECC85s then had the full h.t. available for signal swing, and porportioned that to fit in. That then solved the heater-cathode voltage problem. (I found this 'oversight' in several other amplifiers as well. I had to elevate the heater supply to a suitable voltage to 'fit in'. Some had heaters just floating. How do those folks design? )
Indeed. See my previous notes. As said, I found the anode-cathode voltage(s) for the ECC85 too low to 'fit in' the signal required for the EL34s without 'bottoming'.
Cannot recall now but ..... oh, of course. I changed the circuit to ac-coupled, so that the ECC85s then had the full h.t. available for signal swing, and porportioned that to fit in. That then solved the heater-cathode voltage problem. (I found this 'oversight' in several other amplifiers as well. I had to elevate the heater supply to a suitable voltage to 'fit in'. Some had heaters just floating. How do those folks design?
)- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.