The large electrolytics have long leads serve as much as aerials due to the inductance in the aluminium foil winding. These could be responsible for the anomalies in your scope images. ICs work best when the supply pins have capacitance close to them with a short path to earth as possible for the zero volts end. That is probably what is hidden from the underside of the pcb. One modification of the supply would be to solder a 100n polyester capacitor between the IC supply pins 4 and 11 so glitches due to differences at hf in each rail cancel out. I have benefited from doing this and I have seen somewhere that this approach is one recommended by Self.
I did find connecting the bypass works better when connected to cap pins rather than central earth near the preamp grond. It seemed to solve the main noise discrepancy. I will want to reduce some noise overall but t for now that will do.
That’s an excellent diagram!
What I’d really like would be to know the factory bias.
I have seen two values for the voltage across for 0R22 to give mA. I haven’t touched the pots. The DC offset on the output is <10mV and the channels are close at 2-3mA between them.
only thing issue is i replaced a 6.8K with a 6.81K on both channels.
I did check the resistors to the pre stage power supply They are 220R out of circuit but that doesnt mean that MF didnt update at a later stage.
I have also been eyeing up the EIZZ log 50K pot. However the EIZZ is provided with carbon film resistors so to replace the ALPS pot it would be £50 for the EIZZ and then an extra £50 to replace the resistors within it. That would be better for noise than the carbon wiper on the ALPS pot, however I don't know if it's worth it for this amp.
Also one simple mod I note that MF didn't do is twist the + and - wires from the bridge rectifier to the preamp power supply. As the wires are carrying AC, they're crossing tracks and ground leads. In tube amps twisting the AC heater supplies (high current low voltage) is common practice to reduce the field around them. MF did do this for the LEDs but they're fed using DC so that's more for neatness.
In terms of grounding - the preamp ground travels across tracks at 90 then via a hookup wire to the ground at the relay then takes a path back to the main caps rather than back to the pre-amp capacitor ground pins (the current loop can still do this but it's taking a long path. I'm wondering if passing the ground from the MC33079 direct to the ground of the pre-power supply would help reduce noise.
I think today I want to do the paper and pen sums for the stability, neaten up the mods ready for re-assembling the amp. I may order 100Vdc 1u MKP10 and FKP1 for the coupling as the amp is currently using my 640Vdc caps for tube amp.
Also one simple mod I note that MF didn't do is twist the + and - wires from the bridge rectifier to the preamp power supply. As the wires are carrying AC, they're crossing tracks and ground leads. In tube amps twisting the AC heater supplies (high current low voltage) is common practice to reduce the field around them. MF did do this for the LEDs but they're fed using DC so that's more for neatness.
In terms of grounding - the preamp ground travels across tracks at 90 then via a hookup wire to the ground at the relay then takes a path back to the main caps rather than back to the pre-amp capacitor ground pins (the current loop can still do this but it's taking a long path. I'm wondering if passing the ground from the MC33079 direct to the ground of the pre-power supply would help reduce noise.
I think today I want to do the paper and pen sums for the stability, neaten up the mods ready for re-assembling the amp. I may order 100Vdc 1u MKP10 and FKP1 for the coupling as the amp is currently using my 640Vdc caps for tube amp.
There is an order for taking zero voltage points to ground which should be at the chassis. The dirtiest ones like the transformer centre tap the speaker return should connect first. You don't want to have the phono preamp or the line stage at ground especially the nfb RC connection, where all the crud from such sources flowing through them. Bonsai who is a member of our group has a website you could look at for more detailed comment.
Thank you I'll take a look. I've learnt most of the grounding design from tube amps.
Twisting of the wires from the BR to the preamp/MM power regulated power supply made a difference. It probably helped lifting the cables off the signal source tracks that they cross too. I was wrong they carry DC not AC, however that will help either way! I need another coffee this morning!
Case back together again.
i managed to destroy some of the old pads, so in the end I had to run a shielded cable from the tapeout (ie selected source) tothe input resistors of the preamp bypassing the tape monitor switch.
Good clarity improvement so the source select/volume is really a weak point.
i managed to destroy some of the old pads, so in the end I had to run a shielded cable from the tapeout (ie selected source) tothe input resistors of the preamp bypassing the tape monitor switch.
Good clarity improvement so the source select/volume is really a weak point.
Looks a little backstreet but there is heat shrink under that electrical tape, the caps I need to find a plastic support but for now that will do.
Mounted topside due to the pads disintegrating. You can see the two shielded cable (this has one shielded and the unshielded conductor isn't used). The shield is tied to the grounding link. These take the signal from the tape out (connected direct to the signal path from the selector) and that bypasses the monitor switch.
The 640Vdc MKP10 1uF and 640Vac FKP1 are a little large compared to the Panasonic ECW 😀 The 10uF 100Vdc MKP4 at the front have a nice site to sit.
The 3300uF caps are on the underside, along with the frequency compensation and the Schaffner line filter fits nicely and the boot fits!
Listening last night the amp has more clarity than I think it ever had, it certainly has more grunt but even the bass has more detail and clarity. The initial harsh treble has dropped off and given an hour warmup it still has the same character as the original amp.
For anyone with one of these amps - it does have some upgrade potential within the design limits. After my messing these are my thoughts:
1. Main power, main capacitance and ripple
Ensure your main capacitance has 5.6A ripple capability and the original values have a 1.2Vpk ripple and to remove that you'd need add an additional 15,000uF per channel. Fitting four 15,000uF cans in could be possible but you may want to move the main supply off onto a separate board with a bit of moving of the toroid, adjusting the position of the X2 cap on the switch to fit. It also has the benefit that rectified noise remains off board. However ensure the wires you use are (a) high temp insulation and (b) capable of taking 8A.
The four 3300uF caps simply provide local current loop decoupling but there's no resistance providing a RC filter to the power stage however removing them would increase noise across the PCB underside.
One of the simplest enhancements is simply to add a line filter on the mains. The filter needs to support 6A (IIRC the amp draws 4.5A from the AC mains and 5A may be too close to the mark). Also check the attenuation curves for the noise you get and the back space between the toroid and the filter. The Schaffner I used fitted the original plug hole and even lined up the screw holes with just a squaring of the hole.
2. Pre-amp/MM Power supply Capacitance
I don't think this needs too much of a boost in capacitance but the new Alu Polymer caps fit nicely and provide exceptionally low ESR but they have a horrific derating for low frequency but given their ripple current is 4000mA and the consumption is <50mA this isn't a problem, paralleling two 560uF replaces the 1000uF main reserve with 1120uF but with an ESR of 5mOhm but to be honest I think adding a low ESR reserve closer to the preamp if you don't use the MM section is possibly a better option as it reduces the noise across the PCB.
Also the supply wires that connect the main caps to the supply are laid freely across the board. Simply twisting them together and lifting them off the board slightly helps improve noise.
My focus here was to minimise ripple at any frequency to reduce the noise on the pre and MM stages. More advanced changes could be replacing the regulation and using a more advanced low drop out IC.
Ceramic decoupling in parallel with the existing electrolytic decoupling at the MC33079 is something I would recommend.
3. Improving the signal path.
The reality is that the signal path here can really be improved:
a) If you want to change the sound then you can replace the quad opamp with a more modern opamp however I want to keep with the old.
b) The source selector isn't great and a replacement would be invasive but there's enough links to be able to hook a replacement or even a relay board.
c) The source signal takes a road trip around the PCB, shortening this and using shielding helped the clarity immensely.
d) Taking the tape monitor switch out of the signal path really makes an audible improvement in clarity. The suggestion here is, if a relay board is used you could make the tape monitor switch control a tape monitor relay on the board.
e) The volume pot - well this does sit right beside the furnace-hot heatsink so as the amp heats up so does the Johnson-Nyqusit noise of the carbon sweeper. The pot feeds back into the opamp output side as a divider with the output resistors, so improving the pre-amp resistors needs an improvement on the volume pot for completeness. A mechanical gearing could be used to move the volume pot away and up towards the pre-amp.
Putting a stepped attenuator would help improve clarity but being right next to the heatsink, simply would help reduce noise.
f) some additional cooling holes in the base of the chassis could help reduce the temperature around the preamp section.
g) larger higher grade resistors will work nicely but see point (e)
h) replacing the preamp input caps from electrolytics to metal PP really works nicely.
i) One of the top improvements was replacing the electrolytic coupling cap between the pre-amp and power amp. Your choice here, but increasing the capacitance gives more bass extension BUT ensure you have something quick here to improve detail.
j) I took the approach that bias current flow through the stage transistors (the LTP, the VAS etc) all increase noise if the resistor in series with that flow is operating hot (discoloured). So increasing the wattage helps reduce the operating temperature (given these had discoloured there is obviously an unrated part). This is why the CPF (the orange "DALE" resistors) are either CPF2 2W or CPF3 3W. The additional size gives more area to cool too. Also with the 1% or 0.1% tolerance and low ppm temperature movement helps keep the amp noise stable.
Additional improvements here are the bias resistors and base blocker resistors of the BUZ fets, update of the LTP resistors too. However I changed the resistors that obviously were under-rated rather than all of them. I suspect once you are finished then I suspect replacing the bourns pot with a CPF low ppm resistor would improve the sound further. Given I've not changed the bias in 24 years, these are more trim and forget so why not replace with highly stable lower noise resistor?
k) I've put frequency compensation into the power stage along with ceramic decoupling on the LTP to reduce noise.
Now the amp is together, the next step is to measure the noise and response with the case fully reassembled (offering RF protection). There's some more enhancements (such as replacing some of the resistors like the R220 power supply and the other resistors in the pre-amp stage) but for now it works and sounds better than it ever did without loosing too much of the character of the amp.
Mounted topside due to the pads disintegrating. You can see the two shielded cable (this has one shielded and the unshielded conductor isn't used). The shield is tied to the grounding link. These take the signal from the tape out (connected direct to the signal path from the selector) and that bypasses the monitor switch.
The 640Vdc MKP10 1uF and 640Vac FKP1 are a little large compared to the Panasonic ECW 😀 The 10uF 100Vdc MKP4 at the front have a nice site to sit.
The 3300uF caps are on the underside, along with the frequency compensation and the Schaffner line filter fits nicely and the boot fits!
Listening last night the amp has more clarity than I think it ever had, it certainly has more grunt but even the bass has more detail and clarity. The initial harsh treble has dropped off and given an hour warmup it still has the same character as the original amp.
For anyone with one of these amps - it does have some upgrade potential within the design limits. After my messing these are my thoughts:
1. Main power, main capacitance and ripple
Ensure your main capacitance has 5.6A ripple capability and the original values have a 1.2Vpk ripple and to remove that you'd need add an additional 15,000uF per channel. Fitting four 15,000uF cans in could be possible but you may want to move the main supply off onto a separate board with a bit of moving of the toroid, adjusting the position of the X2 cap on the switch to fit. It also has the benefit that rectified noise remains off board. However ensure the wires you use are (a) high temp insulation and (b) capable of taking 8A.
The four 3300uF caps simply provide local current loop decoupling but there's no resistance providing a RC filter to the power stage however removing them would increase noise across the PCB underside.
One of the simplest enhancements is simply to add a line filter on the mains. The filter needs to support 6A (IIRC the amp draws 4.5A from the AC mains and 5A may be too close to the mark). Also check the attenuation curves for the noise you get and the back space between the toroid and the filter. The Schaffner I used fitted the original plug hole and even lined up the screw holes with just a squaring of the hole.
2. Pre-amp/MM Power supply Capacitance
I don't think this needs too much of a boost in capacitance but the new Alu Polymer caps fit nicely and provide exceptionally low ESR but they have a horrific derating for low frequency but given their ripple current is 4000mA and the consumption is <50mA this isn't a problem, paralleling two 560uF replaces the 1000uF main reserve with 1120uF but with an ESR of 5mOhm but to be honest I think adding a low ESR reserve closer to the preamp if you don't use the MM section is possibly a better option as it reduces the noise across the PCB.
Also the supply wires that connect the main caps to the supply are laid freely across the board. Simply twisting them together and lifting them off the board slightly helps improve noise.
My focus here was to minimise ripple at any frequency to reduce the noise on the pre and MM stages. More advanced changes could be replacing the regulation and using a more advanced low drop out IC.
Ceramic decoupling in parallel with the existing electrolytic decoupling at the MC33079 is something I would recommend.
3. Improving the signal path.
The reality is that the signal path here can really be improved:
a) If you want to change the sound then you can replace the quad opamp with a more modern opamp however I want to keep with the old.
b) The source selector isn't great and a replacement would be invasive but there's enough links to be able to hook a replacement or even a relay board.
c) The source signal takes a road trip around the PCB, shortening this and using shielding helped the clarity immensely.
d) Taking the tape monitor switch out of the signal path really makes an audible improvement in clarity. The suggestion here is, if a relay board is used you could make the tape monitor switch control a tape monitor relay on the board.
e) The volume pot - well this does sit right beside the furnace-hot heatsink so as the amp heats up so does the Johnson-Nyqusit noise of the carbon sweeper. The pot feeds back into the opamp output side as a divider with the output resistors, so improving the pre-amp resistors needs an improvement on the volume pot for completeness. A mechanical gearing could be used to move the volume pot away and up towards the pre-amp.
Putting a stepped attenuator would help improve clarity but being right next to the heatsink, simply would help reduce noise.
f) some additional cooling holes in the base of the chassis could help reduce the temperature around the preamp section.
g) larger higher grade resistors will work nicely but see point (e)
h) replacing the preamp input caps from electrolytics to metal PP really works nicely.
i) One of the top improvements was replacing the electrolytic coupling cap between the pre-amp and power amp. Your choice here, but increasing the capacitance gives more bass extension BUT ensure you have something quick here to improve detail.
j) I took the approach that bias current flow through the stage transistors (the LTP, the VAS etc) all increase noise if the resistor in series with that flow is operating hot (discoloured). So increasing the wattage helps reduce the operating temperature (given these had discoloured there is obviously an unrated part). This is why the CPF (the orange "DALE" resistors) are either CPF2 2W or CPF3 3W. The additional size gives more area to cool too. Also with the 1% or 0.1% tolerance and low ppm temperature movement helps keep the amp noise stable.
Additional improvements here are the bias resistors and base blocker resistors of the BUZ fets, update of the LTP resistors too. However I changed the resistors that obviously were under-rated rather than all of them. I suspect once you are finished then I suspect replacing the bourns pot with a CPF low ppm resistor would improve the sound further. Given I've not changed the bias in 24 years, these are more trim and forget so why not replace with highly stable lower noise resistor?
k) I've put frequency compensation into the power stage along with ceramic decoupling on the LTP to reduce noise.
Now the amp is together, the next step is to measure the noise and response with the case fully reassembled (offering RF protection). There's some more enhancements (such as replacing some of the resistors like the R220 power supply and the other resistors in the pre-amp stage) but for now it works and sounds better than it ever did without loosing too much of the character of the amp.
I finally managed to have a look at the biasing difference between the two channels. The amp had been on with the cover for an hour along with the scope before a self calibration. 8R load resistors.
First I managed to scare myself stupid as one channel suddenly dropped by 1/2 amplitude.. Turns out that the selector switch needs a careful "wobbling" on the odd occasion so I think that really needs some cleaner/deoxy spray.
The scope was showing this before hand:
And this after:
With the scope setup measuring the 0R22 on the left channel that doesn't clip, I knew the left channel was about 95mV across the 0R22 resistor. Trying to get the same value on the right channel was difficult switching from 80mV to 100mV however I managed to get it closer to the left channel. There's now 6mV difference (27mA) difference between the two channels at full volume. 95mV means 431mA. So a 6% difference. Trying to trim it is like Skywalker trying to hit the exhaust vent on the deathstar.
That will do. Before I discover yet more design 'goals' with the amp (ie the selector).
Interestingly there's some high spikes:
First I managed to scare myself stupid as one channel suddenly dropped by 1/2 amplitude.. Turns out that the selector switch needs a careful "wobbling" on the odd occasion so I think that really needs some cleaner/deoxy spray.
The scope was showing this before hand:
And this after:
With the scope setup measuring the 0R22 on the left channel that doesn't clip, I knew the left channel was about 95mV across the 0R22 resistor. Trying to get the same value on the right channel was difficult switching from 80mV to 100mV however I managed to get it closer to the left channel. There's now 6mV difference (27mA) difference between the two channels at full volume. 95mV means 431mA. So a 6% difference. Trying to trim it is like Skywalker trying to hit the exhaust vent on the deathstar.
That will do. Before I discover yet more design 'goals' with the amp (ie the selector).
Interestingly there's some high spikes:
Hi @NickKUK, I had downloaded your spice files the other day as I am likely to buy a used A2 and wanted to "prepare" a bit.
The original simulations with the BUZ's you had included worked quite OK, but when I changed the models to the still available Exicon ECX10N20 and ECX10P20, I was not able to bias at all.
And after some playing and staring, I found that you seem to have drain and source the wrong way for the P FETs.
@jaycee has it right in https://www.diyaudio.com/community/...-output-mosfets-installed.383382/post-7011287 I think.
After fixing that, I was able to get it running properly, but needed to add 10 nF caps between gate and source of the FTEs to prevent a nasty 7 MHz oszillation. That BTW also happened already after changing the BUZ906 to their correct direction.
@jaycee I am not sure I understand your comment about the open loop gain stability. The amp has feedback, so in which way does that tell anything? I am an educated noob only, so bear with me 🙂 if that question is "dumb".
Thanks,
Mo
The original simulations with the BUZ's you had included worked quite OK, but when I changed the models to the still available Exicon ECX10N20 and ECX10P20, I was not able to bias at all.
And after some playing and staring, I found that you seem to have drain and source the wrong way for the P FETs.
@jaycee has it right in https://www.diyaudio.com/community/...-output-mosfets-installed.383382/post-7011287 I think.
After fixing that, I was able to get it running properly, but needed to add 10 nF caps between gate and source of the FTEs to prevent a nasty 7 MHz oszillation. That BTW also happened already after changing the BUZ906 to their correct direction.
@jaycee I am not sure I understand your comment about the open loop gain stability. The amp has feedback, so in which way does that tell anything? I am an educated noob only, so bear with me 🙂 if that question is "dumb".
Thanks,
Mo
Last edited:
When I open that file, I get a missing symbols error, despite copying the .sub files to my C:\Program Files\LTC\LTspiceXVII\lib\sub folder. What have I missed?
It is the same for me when opening Jaycee's spice file. Open it anyway. The output MOSFTs will be missing. Just add them using "Draft-Component".
It should look similar to the below afterwards.
Make sure you connect source and drain correctly, like I have in the image. In some spice files in this thread the P-channels were wrong.