The gain should be less than 1, but I have no idea why it might be greater than 1. I see from your two graphs that the signal level is the same. Did you adjust the input signal to get identical signal levels.
Do you have individual power supplies for each channel? Are they switching supplies? If you have individual supplies, did you try changing the power supply with one from a good amp to see if the problem is with the supply? Did you also try changing the PS board board with one from a good amp to see if the problem is with the PS board?
The increase in noise in both the good and bad amps at about 9kHz is not normal either.
Pictures of your amps would be helpful.
Arrangement of cables, power cords, proximity to power transformers, etc, and arrangement of amplifier and test apparatus may also affect the test measurements. Photos of that may be helpful too.
Do you have individual power supplies for each channel? Are they switching supplies? If you have individual supplies, did you try changing the power supply with one from a good amp to see if the problem is with the supply? Did you also try changing the PS board board with one from a good amp to see if the problem is with the PS board?
The increase in noise in both the good and bad amps at about 9kHz is not normal either.
Pictures of your amps would be helpful.
Arrangement of cables, power cords, proximity to power transformers, etc, and arrangement of amplifier and test apparatus may also affect the test measurements. Photos of that may be helpful too.
Ben,
Thanks for your reply. I just did a sanity check on the gain level using 1kHz sine wave and multimeter. It is <1 in all cases. My digital measurement was giving me the wrong gain factor.
There is no transformer in the chassis. It is using the same Meanwell 36V switching power supply as the project specified. I thought the noise at 9kHz is probably from the switching power supply, though I'm surprised that the LT1084 didn't do any attenuation to it. I thought the noise level is low enough that I wasn't too concern about it.
I have attached a picture of the build. The power supply board basically provide some RC and LC filters, and the relay output muting function. The R is currently bypassed in order the add the voltage regulator inline.
You may notice that I'm using a much smaller output coupling capacitor than the 10,000uF specified as I'm using them as high pass filter to drive the tweeter directly.
Thanks for your reply. I just did a sanity check on the gain level using 1kHz sine wave and multimeter. It is <1 in all cases. My digital measurement was giving me the wrong gain factor.
There is no transformer in the chassis. It is using the same Meanwell 36V switching power supply as the project specified. I thought the noise at 9kHz is probably from the switching power supply, though I'm surprised that the LT1084 didn't do any attenuation to it. I thought the noise level is low enough that I wasn't too concern about it.
I have attached a picture of the build. The power supply board basically provide some RC and LC filters, and the relay output muting function. The R is currently bypassed in order the add the voltage regulator inline.
You may notice that I'm using a much smaller output coupling capacitor than the 10,000uF specified as I'm using them as high pass filter to drive the tweeter directly.
Attachments
You previously mentioned four channels with one channel having high noise. Did you try swapping power supplies to try to track down the source of the noise?
You just mentioned that you bypassed the R of the RC filter and added a voltage regulator. I do not know exactly what you have done. A schematic would be helpful. My suspicion is that this change may have caused the noise issues. By bypassing the R, you have removed a stage of filtering. If you want to add a regulator, you should leave the power supply as designed and add the regulator after the RC stage.
I suggest you remove the voltage regulator and build the power supply as designed. Then listen and test. If it then tests noisy, there may be an issue with your build since this is a tested design.
After you get the amps to meet the design specifications, then you can modify them if you still have the desire to do more.
You just mentioned that you bypassed the R of the RC filter and added a voltage regulator. I do not know exactly what you have done. A schematic would be helpful. My suspicion is that this change may have caused the noise issues. By bypassing the R, you have removed a stage of filtering. If you want to add a regulator, you should leave the power supply as designed and add the regulator after the RC stage.
I suggest you remove the voltage regulator and build the power supply as designed. Then listen and test. If it then tests noisy, there may be an issue with your build since this is a tested design.
After you get the amps to meet the design specifications, then you can modify them if you still have the desire to do more.
I built 4 channels in two stereo chassis. Each stereo chassis use one switching power supply. I started out with the passive RC+LC filter on the power supply and found that I have a 120Hz hum/buzz coming through my speakers (since they have 102dB and 107dB/W sensitivity). So I proceeded to add the LT1084 to reduce the low frequency noise. Before adding the voltage regulator, I have a similar noise floor rise with the passive power supply on the bad channel, but with higher 60/120 Hz components and harmonics.
Today, I tried swapping power supply between the two channels and the noise characteristics changed, but not reduced. So I try to add another ground wire twisted on the V+ wire since currently there is a section of 5 inches that is not twisted. That seem to solve the problem.
I don't quite understand why the other channel with similar layout doesn't show the problem, but I will add the ground wire to it as preventive measure. May be shielded power wiring could further bring down the noise floor?
As to why this could solve the problem, I'm not quite sure. A wire picking up interference on specific frequencies from a nearby radiation source is understandable. A broadband noise floor like that is something that I have not seen before.
For the 9kHz noise band, it may be worth further investigation. But probably after I spend some time enjoying listening to the amp first.
Thanks for your suggestion.
Today, I tried swapping power supply between the two channels and the noise characteristics changed, but not reduced. So I try to add another ground wire twisted on the V+ wire since currently there is a section of 5 inches that is not twisted. That seem to solve the problem.
I don't quite understand why the other channel with similar layout doesn't show the problem, but I will add the ground wire to it as preventive measure. May be shielded power wiring could further bring down the noise floor?
As to why this could solve the problem, I'm not quite sure. A wire picking up interference on specific frequencies from a nearby radiation source is understandable. A broadband noise floor like that is something that I have not seen before.
For the 9kHz noise band, it may be worth further investigation. But probably after I spend some time enjoying listening to the amp first.
Thanks for your suggestion.
If you have extremely sensitive speakers, you need to pay attention to all factors that may allow noise into your system. I have 103dB sensitive speakers and I have experienced noise, so I have learned to build noise free amps and preamps. For me, noise abatement starts at the beginning of the build, not at the end.
Single ended circuits are especially prone to noise, and the N channel circuits with the power supply connected the drain are even more susceptible to noise. So very clean power is needed. Good grounding techniques and keeping loop areas to the absolute minimum is required. In your case pairing a ground wire with the V+ wire is a case where you reduced a loop area.
In my amps and preamps, all conductors (whether power or signal) and their returns are twisted together. So Ground and V+ together, (Ground and V+ and V- together if bipolar supply), input signal+ and Ground together, output signal or speaker and Ground together.
Keeping AC and power transformers away from the amplifier circuit is also critical. You are using a SMPS so you have no AC in your box, but the SMPS should be kept away from the amplifier. Also helpful is to plug all of your music system components into one wall AC outlet, and as much as physically feasible, bundle all of the AC cords together. That all helps to minimize loop area and noise.
Here is some good information by diyAudio member Bonsai:
https://hifisonix.com/wordpress/wp-content/uploads/2019/02/Ground-Loops.pdf
Single ended circuits are especially prone to noise, and the N channel circuits with the power supply connected the drain are even more susceptible to noise. So very clean power is needed. Good grounding techniques and keeping loop areas to the absolute minimum is required. In your case pairing a ground wire with the V+ wire is a case where you reduced a loop area.
In my amps and preamps, all conductors (whether power or signal) and their returns are twisted together. So Ground and V+ together, (Ground and V+ and V- together if bipolar supply), input signal+ and Ground together, output signal or speaker and Ground together.
Keeping AC and power transformers away from the amplifier circuit is also critical. You are using a SMPS so you have no AC in your box, but the SMPS should be kept away from the amplifier. Also helpful is to plug all of your music system components into one wall AC outlet, and as much as physically feasible, bundle all of the AC cords together. That all helps to minimize loop area and noise.
Here is some good information by diyAudio member Bonsai:
https://hifisonix.com/wordpress/wp-content/uploads/2019/02/Ground-Loops.pdf
I have one channel that picks up noise; I can only hear it with my sensitive 32 ohm headphones. There is a 100 ohm on the gate - and the leads to the resistors have a 'shield' with a twisted earth wire.
Both connection wires were suspended in mid air. Then I found out that the wire of noisy channel improved when I changed the distance to the metal of the chassis - closer, it started to disappear. The wire being a bit short, I could not reroute it exact;y hidden in the protection of the heatsink. I should get a coax cable. Also this movement was out of the way of the small transform er I have for the -25V bias voltage - it could be stray magnetics or static.
Thinking about that is on the Heavy Side for me. (pun intended).
Both connection wires were suspended in mid air. Then I found out that the wire of noisy channel improved when I changed the distance to the metal of the chassis - closer, it started to disappear. The wire being a bit short, I could not reroute it exact;y hidden in the protection of the heatsink. I should get a coax cable. Also this movement was out of the way of the small transform er I have for the -25V bias voltage - it could be stray magnetics or static.
Thinking about that is on the Heavy Side for me. (pun intended).
I have implemented the folded PCC88 driver. The first model was with a standard CCS from LT Spice. Actually I used a JFET for it, J103, giving Idss of 20 ma, with a 39 ohm, it is 15 mA. That is split across the tubes (2,6 mA) and the choke.
Now the follower is a compound one, not a darlington (two PNP) but a Darlingnot (a PNP and an NPN; in fact I choose a high voltage P-FET that can handle 50Vds and a BD139). I choose that because in modelling the standard darlington gave 5th and 7th harmonics; while the darlingnot has a natural degradation. So not the 'ideal' no distortion.
Measured: A trace of second, at 0,03% about and 8VRMS out. All I need. But - repeat: this circuit can give a clean 40Vpp from 40V Vb!
Al together, not bad.
BW from 6Hz to 105 kHz. Two halves of the transformer parallel. Each has Rdc of 1k2; parallel it is 600 ohms. Parallel it behave nice.
There are a few minor bumps of about 2 dB at the higher end. The BW fall gracefully.
Without current just on the signal generator the 'choke' had a higher BW. The 'choke' is a 10K:10K transformer. It was sold to me as allowing 45mA DC. I doubt the specs, I somehow distrust it to be a PP transformer not SE as they said.
Channels are within less than 0,1 dB.I'm finally getting back to the Sony chassis TA 4650 build after three non functioning power supplies. Got the wall wart 36v from DIY store, works, tested. I know the cards and transistors work great because they were bench tested. What I have now is logistics and manipulation to get things in the chassis, which I think is quite do'able now.
Question on wall wart power supply: There are two output rails of V+ and V-. Does this mean if these two branches are used, each rail has 36v with half the current if they are both used at the same time i.e. one branch for each card? If only one branch is used, will that provide 36v at full current that can be split between the cards with just two wires output from the wall wart?
Also, the
Question on wall wart power supply: There are two output rails of V+ and V-. Does this mean if these two branches are used, each rail has 36v with half the current if they are both used at the same time i.e. one branch for each card? If only one branch is used, will that provide 36v at full current that can be split between the cards with just two wires output from the wall wart?
Also, the
SMPS DC FILTER P089ZB
that I have constructed has parameters of 48V and 3A. Can I use a single one of these for both channel boards @36V? Or do I need a single card for each channel separately?
Can't comment on your build as I don't know how much current it draws.
However, on Papa's 1st lottery VFET, I successfully used 2 of Mark's excellent filters, that is 1 per channel.
Just for information:
All these combos were tried with short cables and connections enabling very quick changes between configurations and blind listening thanks to the other operator.
I would go for first with 1 filter per channel, that is 2 in total in your case, provided you are still well within the amp rating of the filters. Then of course you may want to experiment. My bet is also that separate filters for each channel help somewhat separating the channel PS... perhaps.
I hope this helps
Claude
However, on Papa's 1st lottery VFET, I successfully used 2 of Mark's excellent filters, that is 1 per channel.
Just for information:
- I tried with one for both channels, something I did well knowing I was driving very slightly overspecced and something I would NOT recommend other than for testing purposes when you know what you are doing. I also knew well what components could withstand in real life for this test. Bottom line, it sounded better and but less so than with 2 filters (1 per channel).
- 2 filtersn, that is 1 filter per channel, sounded best and was well within the filter's specs. It is not the first time I noticed that these excellent filters sound at their best when well within their specs, which is of course no wonder, regardless the (higher) ampere rating of their components etc.
- I tried 4 filters, that is 2 per channel, cascading them. Sounded slightly worst than 1 per channel to my ears and the other listener involved. Of course that is very amp dependant.
All these combos were tried with short cables and connections enabling very quick changes between configurations and blind listening thanks to the other operator.
I would go for first with 1 filter per channel, that is 2 in total in your case, provided you are still well within the amp rating of the filters. Then of course you may want to experiment. My bet is also that separate filters for each channel help somewhat separating the channel PS... perhaps.
I hope this helps
Claude
That would mean you are even on the safe side with only one filter that can be conveniently placed, looped in or not, between PS brick and amp. Ideal for quick testing purposes... and os eay to implement long time.
Doesn't mean though - if you like it - that you wouldn't want to try more of these, but at least thanks to ZM it appears your build can get started with only one filter, great news
Claude
Doesn't mean though - if you like it - that you wouldn't want to try more of these, but at least thanks to ZM it appears your build can get started with only one filter, great news
Claude
Thanks so much to mighty ZM an Claude. However, I am still wondering if just one of the V+ V- rails on the plug carry the full current load, and if each of the V+V- carry 36v with half the current if both are used at the same time. It's hard to get a lead in that tiny plug to measure anything.
