One problem you may have with the bulb limiter is that there is no local decoupling on the Vcc rail on the PCB. With MJ15003's you may have a greater tendency to oscillate than with the original 2N3055's. if you have a spare 100uF 100V cap you could connect that across the supply when testing. That may help to suppress oscillations.
At this point I wonder whether another rebuild on the L channel board using new parts is needed. Does the R board work with the left channel output trannys? If so you could reuse those on the L channel while things are being sorted.
I would be inclined to use all new resistors as well as caps and trannys. If you wish to keep the bias diodes I suggest checking the forward voltage of any 1N4148's and if more than about 0.65 at 3mA try using 1n4448's. Or even diode connected BC547's.
Are you sure that the output capacitor isn't shorted? Or if R1L isn't low?
(I think if the faulty component(s) are identified the board should work with MJ15003's, but I've not actually tried that. I've successfuly used epi 3055's (with mods) but the 15003 ft is between the two (38494 = 2N3055H (I assume) and is 800kHz , epi 3055s are 2.5MHz but MJ15003 is 2MHz. If your 38494's are still alive that is helpful.
At this point I wonder whether another rebuild on the L channel board using new parts is needed. Does the R board work with the left channel output trannys? If so you could reuse those on the L channel while things are being sorted.
I would be inclined to use all new resistors as well as caps and trannys. If you wish to keep the bias diodes I suggest checking the forward voltage of any 1N4148's and if more than about 0.65 at 3mA try using 1n4448's. Or even diode connected BC547's.
Are you sure that the output capacitor isn't shorted? Or if R1L isn't low?
(I think if the faulty component(s) are identified the board should work with MJ15003's, but I've not actually tried that. I've successfuly used epi 3055's (with mods) but the 15003 ft is between the two (38494 = 2N3055H (I assume) and is 800kHz , epi 3055s are 2.5MHz but MJ15003 is 2MHz. If your 38494's are still alive that is helpful.
Last edited:
Good question - I changed them out because we thought the -6V clipping was due to TR2L
Yes because that same capacitor works with the R board.
So OK then...
Another random thought on the stability question...
Are the 0.3 ohms original, or have they been replaced with possibly inductive wire wounds in the past?
Also, if you have the outputs on flying leads to the heatsink then are we compounding any issues due to wiring inductance... which could trigger instability.
Another random thought on the stability question...
Are the 0.3 ohms original, or have they been replaced with possibly inductive wire wounds in the past?
Also, if you have the outputs on flying leads to the heatsink then are we compounding any issues due to wiring inductance... which could trigger instability.
So it turns out that what we thought was a duff lower OP device from L channel, which we thought might be causing the -6V clipping actually turns out to be OK when connected to R channel board - both of the OP devices from L channel work perfectly!!!
They are original ones - green 3W AWC resistors - although the component list says they are 3W AMC!!
I think John is right - L board needs a total rebuild.
Probably got the resistors hanging around - and also most mof the caps (Wima) -
Thinking the following:
TR100 - BC558
TR101 - BC109
TR102 - BC546
TR103 - BC546
TR104 - BC556
TR105 - 2N5322
TR106 - 2N5320
OUTPUTS- MJ15003 - or leave the 38494s?
They are original ones - green 3W AWC resistors - although the component list says they are 3W AMC!!
TBH that did cross my mind earlier - so I made the leads flying leads shorter than the ones inside the amp.
I think John is right - L board needs a total rebuild.
Probably got the resistors hanging around - and also most mof the caps (Wima) -
Thinking the following:
TR100 - BC558
TR101 - BC109
TR102 - BC546
TR103 - BC546
TR104 - BC556
TR105 - 2N5322
TR106 - 2N5320
OUTPUTS- MJ15003 - or leave the 38494s?
I'd recommend a slightly different line-up.
Tr100- BC559B but a BC558B should be OK (BC154 was low noise/high gain)
Tr101- Yes if you have a BC109 or BC549 if not
Tr102, Tr103 - I think the BC546 is marginal (65V). Maybe ok but I'd suggest MPSA06 or 2N5551 as a safer option
Tr104 - MPSA56 or 2N5401 (BC556 may work, but ..)
Tr105 -2N5322 should be OK (this is the PNP). Alternatives 2N4033, 2N5680, (BD140 with boar layout change)
Tr106 -2N5320 should be OK (this is the NPN). Alternatives 2N3019, 2N5682 (BD139 with board layout change)
I'd try 38494's if you are happy that they work, first. Can check MJ15003's if board OK.
Repeat warnings- as I and ejp posted, really check the PCB tracks properly. Even looking for low ohms (<1) along the tracks and insulation (>10M or whatever your meter reaches) between tracks. If the amp has had abuse, tracks can be open or short and solder bridges covered in flux. High resistance values in this circuit means some leakage would not be acceptable. You never know. A proper inspection might reveal the missing piece.
Does Quad sell replacement PCB's? I'd think about that, if it were me. Of course you'd need to get caps which fit the locations. Most modern caps are axial but smaller than earlier types, so it should be possible to squeeze non-original spec parts into the board.
One more point - the small 3.3pF capacitor on Tr102 collector is important for stability. I tend to use silver mica for small caps (for stability and voltage) but they're largish and expensive now. You'd need a cap with breakdown voltage >100V. Many small ceramics are only 50V.
Tr100- BC559B but a BC558B should be OK (BC154 was low noise/high gain)
Tr101- Yes if you have a BC109 or BC549 if not
Tr102, Tr103 - I think the BC546 is marginal (65V). Maybe ok but I'd suggest MPSA06 or 2N5551 as a safer option
Tr104 - MPSA56 or 2N5401 (BC556 may work, but ..)
Tr105 -2N5322 should be OK (this is the PNP). Alternatives 2N4033, 2N5680, (BD140 with boar layout change)
Tr106 -2N5320 should be OK (this is the NPN). Alternatives 2N3019, 2N5682 (BD139 with board layout change)
I'd try 38494's if you are happy that they work, first. Can check MJ15003's if board OK.
Repeat warnings- as I and ejp posted, really check the PCB tracks properly. Even looking for low ohms (<1) along the tracks and insulation (>10M or whatever your meter reaches) between tracks. If the amp has had abuse, tracks can be open or short and solder bridges covered in flux. High resistance values in this circuit means some leakage would not be acceptable. You never know. A proper inspection might reveal the missing piece.
Does Quad sell replacement PCB's? I'd think about that, if it were me. Of course you'd need to get caps which fit the locations. Most modern caps are axial but smaller than earlier types, so it should be possible to squeeze non-original spec parts into the board.
One more point - the small 3.3pF capacitor on Tr102 collector is important for stability. I tend to use silver mica for small caps (for stability and voltage) but they're largish and expensive now. You'd need a cap with breakdown voltage >100V. Many small ceramics are only 50V.
Last edited:
Just checked the BC556 datasheet. Bvcbo is 80V so maybe not as critical as I thought. Curiously ON Semi seem to have obsoleted the MPsA56 but not the MPSA06. You may get lower distortion with a higher voltage VAS transistor. So it may be worth MPSA06 int he VAS stage even if you decide to use BC546/BC556 drivers
Yeah, I was struggling to find this, only place that had them was Cricklewood Electronics, I'll order MPSA06.
All components now delivered - will start rebuilding board soon - need to do a bit of remedial work on the PCB, a couple of pads have detached.
I removed all components from the board and checked each one as I went - all seem to measure out ok, so maybe a cumulative thing. Unfortunately I seem to have broken the leg off L100 - all the paperwork I have does not mention the value, I'm 'presuming' its around 0.01mH -not sure of the DCR though.....
I removed all components from the board and checked each one as I went - all seem to measure out ok, so maybe a cumulative thing. Unfortunately I seem to have broken the leg off L100 - all the paperwork I have does not mention the value, I'm 'presuming' its around 0.01mH -not sure of the DCR though.....
It will probably be around 4 Micro Henry give or take and so 0.004 mH. It must be air cored, no magnetic core or former and the DC resistance will automatically be low because of the thick wire used.
In practice about 10 or twelve turns of 1mm wire I would guess. Pull the old one apart... it may not be all that substantial given the modest power output.
In practice about 10 or twelve turns of 1mm wire I would guess. Pull the old one apart... it may not be all that substantial given the modest power output.
Many old RCA circuits which used 2N3055's (old H type) suggested 10 uH.
I agree that it should be air cored, but I'm curious as the illustrations I've seen on the Q303 suggest it is quite small. Does your component have a ferrite core?
I'm not sure how well the Quad 303 worked with electrostatic speakers. I vaguely recall Quad offering some sort of component (inductor?) to add externally with those. Maybe a larger inductor helps with ELS speakers.
If you do not plan to use els speakers you can probably use almost any coil in the output. Typically it was as Mooly suggested about a dozen turns - usually wound on a 2W 10 ohm resistor. In the old days they may have been bigger than they are now, though. A typical carbon 2W measured 8x17.5mm (just happened to have on in front of me) and with about 20 turns of 0.7mm (22swg in old money) in one layer along the resistor, as I used, it still only works out to be around 1.25uH. That does not impact the audio band significantly. You could probably use an old Bic pen tube as a former. As long as it does not oeverheat! Resistance should be 24m so only dissipates 150mW at full power.
I agree that it should be air cored, but I'm curious as the illustrations I've seen on the Q303 suggest it is quite small. Does your component have a ferrite core?
I'm not sure how well the Quad 303 worked with electrostatic speakers. I vaguely recall Quad offering some sort of component (inductor?) to add externally with those. Maybe a larger inductor helps with ELS speakers.
If you do not plan to use els speakers you can probably use almost any coil in the output. Typically it was as Mooly suggested about a dozen turns - usually wound on a 2W 10 ohm resistor. In the old days they may have been bigger than they are now, though. A typical carbon 2W measured 8x17.5mm (just happened to have on in front of me) and with about 20 turns of 0.7mm (22swg in old money) in one layer along the resistor, as I used, it still only works out to be around 1.25uH. That does not impact the audio band significantly. You could probably use an old Bic pen tube as a former. As long as it does not oeverheat! Resistance should be 24m so only dissipates 150mW at full power.
Last edited:
Regarding your old components, they may "measure" OK but electrolytics can deteriorate.
You'd really need to measure the distortion and ESL/ESR to be sure. I never built Cyril Bateman's distortion kit, but I have attempted ESL/ESR to find dead caps.
If you are not absolutely sure of old parts, it is better to use all new ones. As long as these are sourced from reliable suppliers they should be pukka.
You'd really need to measure the distortion and ESL/ESR to be sure. I never built Cyril Bateman's distortion kit, but I have attempted ESL/ESR to find dead caps.
If you are not absolutely sure of old parts, it is better to use all new ones. As long as these are sourced from reliable suppliers they should be pukka.
The coil is wound on a ferrite core which is 16mm x 6mm - 28 turns of wire on it.
Now that I have access to the other board, the coil on that one measures 0.01mH with a DCR of 0.14 Ohms.
I'm not using ELS, just a pair of Yamaha NS-10M Studio Monitors so as you say, it probably doesn't matter.
I'll have a go at making a couple so at least both channels are the same!!
Now that I have access to the other board, the coil on that one measures 0.01mH with a DCR of 0.14 Ohms.
I'm not using ELS, just a pair of Yamaha NS-10M Studio Monitors so as you say, it probably doesn't matter.
I'll have a go at making a couple so at least both channels are the same!!
Wow... they are making sure of stability with something like that. Ferrite cored output coils would never be used these days.
Is it an acknowledgment of the precarious stability of the output triples I wonder.
It will not affect any rebuild, at least not unless adverse loads are connected.
Is it an acknowledgment of the precarious stability of the output triples I wonder.
It will not affect any rebuild, at least not unless adverse loads are connected.
I think 10uH is for two reasons - a largish capacitive load (like a Quad ELS speaker) and also because the output transistors are somewhat slow. So a small inductor would not provide much assistance for slow transistors.
Ferrite cores would not be used now, but even with a core, the magnetisiing field is not particularly high because the magnetic path is still mostly air. So distortion should not be great. I'd probably just rewind the broken coil with new wire. A 10uH air coil will be quite a bit larger.
Ferrite cores would not be used now, but even with a core, the magnetisiing field is not particularly high because the magnetic path is still mostly air. So distortion should not be great. I'd probably just rewind the broken coil with new wire. A 10uH air coil will be quite a bit larger.
Well that took longer than expected!!
L board now back together with all new components.
Tr100-BC558B
Tr101-BC109
Tr102, Tr103 - 2N5551
Tr104 -2N5401
Tr105 -2N5322
Tr106 -2N5320
T1L-T2L 38494
MR103-104-105-106 1N4148
Rail and mid rail voltages all correct - but still no quiescent adjust - sits around .05mV across emitter resistors.
Swapped out 38494s for MJ15003s - still no quiescent.
Changed the 1N4148s for the original 1S920s and everything sprang into life!!
Emitter resistor voltage now sits stable around 7mV - I'm guessing here that its the characteristics of MR103/104 that is causing this.
I've ordered 1N4448s to try
L board now back together with all new components.
Tr100-BC558B
Tr101-BC109
Tr102, Tr103 - 2N5551
Tr104 -2N5401
Tr105 -2N5322
Tr106 -2N5320
T1L-T2L 38494
MR103-104-105-106 1N4148
Rail and mid rail voltages all correct - but still no quiescent adjust - sits around .05mV across emitter resistors.
Swapped out 38494s for MJ15003s - still no quiescent.
Changed the 1N4148s for the original 1S920s and everything sprang into life!!
Emitter resistor voltage now sits stable around 7mV - I'm guessing here that its the characteristics of MR103/104 that is causing this.
I've ordered 1N4448s to try
Indeed you did - 😱
I could understand if I was just changing these and keeping all the other semiconductors original, I thought It would have been ok considering the 'triple' has been changed to newer devices - and what is interesting there are companies ( and individuals) out there who state that 1n4148s are ok on a total component change.
I'm going to get the other board up to spec first then I'll look closer into this - but for now, I have 1/2 a working 303!!!
thanks to all!!