Oh cool! I just came up with it, didn't realize it was a radical invention... Maybe I should record myself talking in my sleep to see if there are more surprises in store? 😀
I suspect the filtering has caused >20MHz oscillation, since everything is slightly hotter than usual, sounds shot and pressing the "20MHz BW limit" button reduces the filtered rail trace width by half. I will try to root this out tomorrow.
- keantoken
I suspect the filtering has caused >20MHz oscillation, since everything is slightly hotter than usual, sounds shot and pressing the "20MHz BW limit" button reduces the filtered rail trace width by half. I will try to root this out tomorrow.
- keantoken
Okay, today I rediscovered some things I should have been doing all along.
I discovered the amp was not actually oscillating. There is an odd 100MHz sine imposed on the filtered rail voltage, but it doesn't change no matter where I put my fingers, or what the output voltage. It is barely visible, so I don't think it is a problem.
Second, I decided to remove the input filter. Despite originally using a 82R/250p filter, way smaller than in Paul's schematic, it starts to really matter at low volumes when the input pot's impedance is 10-50k depending on the pot you have. A 10k pot gives a -3db point of 85KHz. This limits the effectiveness of square wave tests.
So I moved the 82R/250p filter to right after the RCA jack, as per Bob Cordell's advice to put such a filter there to properly terminate input cables so they don't resonate at >20MHz. Being before the volume pot, changing the volume won't modulate the LP filter. If your amp is sensitive to it's input impedance, this can be important. At high volumes when there is little impedance between the cable and your amp input, the cable combined with an amp's wonky input impedance can become an oscillator. This is far more likely if your input is not far enough away from the VAS, since parasitics here can destabilize the amp.
I also discovered grounding the heatsink greatly helped stability. After the rails filtering, filter removal and heatsink grounding, the amp is faster, and I think I will need to add some miller compensation to make sure. If only I had some way of telling the phase difference of RF signals, I maybe could tune the amp to be stable into all capacitors, without the need for an output inductor.
- keantoken
I discovered the amp was not actually oscillating. There is an odd 100MHz sine imposed on the filtered rail voltage, but it doesn't change no matter where I put my fingers, or what the output voltage. It is barely visible, so I don't think it is a problem.
Second, I decided to remove the input filter. Despite originally using a 82R/250p filter, way smaller than in Paul's schematic, it starts to really matter at low volumes when the input pot's impedance is 10-50k depending on the pot you have. A 10k pot gives a -3db point of 85KHz. This limits the effectiveness of square wave tests.
So I moved the 82R/250p filter to right after the RCA jack, as per Bob Cordell's advice to put such a filter there to properly terminate input cables so they don't resonate at >20MHz. Being before the volume pot, changing the volume won't modulate the LP filter. If your amp is sensitive to it's input impedance, this can be important. At high volumes when there is little impedance between the cable and your amp input, the cable combined with an amp's wonky input impedance can become an oscillator. This is far more likely if your input is not far enough away from the VAS, since parasitics here can destabilize the amp.
I also discovered grounding the heatsink greatly helped stability. After the rails filtering, filter removal and heatsink grounding, the amp is faster, and I think I will need to add some miller compensation to make sure. If only I had some way of telling the phase difference of RF signals, I maybe could tune the amp to be stable into all capacitors, without the need for an output inductor.
- keantoken
If the pot is fed by a voltage source its maximum output impedence is 1/4 of its total value at the setting of 50% of its total value, 2.5k max in the case of a 10k pot
Right, but it's not pot output impedance that matters here, it is impedance from source to output. If the pot is at low volume, there will be say 9.5k between the source and amp input, and this will be in series with the 220p capacitor, forming an LP filter somewhere around 100KHz.
Some kinds of amp input troubles will occur in different ways. Sometimes the amp will oscillate with the pot either at max or at zero. Other times it will oscillate only in middle positions. This goes along with other times where it only oscillates at high or low positions, which indicates interaction with the source/cable or possibly grounding problems.
- keantoken
Some kinds of amp input troubles will occur in different ways. Sometimes the amp will oscillate with the pot either at max or at zero. Other times it will oscillate only in middle positions. This goes along with other times where it only oscillates at high or low positions, which indicates interaction with the source/cable or possibly grounding problems.
- keantoken
My soldering iron tip has slowly eroded down to a tiny spike that bends over if I put too much pressure on it. Is this supposed to happen?
- keantoken
- keantoken
You should replace the tip at first sign of erosion. Once it starts to erode, it can be gone in a matter of days.
bites in bits
It depends on the bit quality. Cheap $10 hobby irons use plain copper rod with thin Nickel plating and erode slowly down to a wire from day 1. Seriously, you can file these to any shape and keep using them until nothing is left, if you wish.
Expensive bits, like those on temp. controlled irons, are usually thick plated with iron and can begin to erode suddenly when the core is exposed, depending on the solder alloy, flux and plating quality. Tin/lead Solders with ~ 0.5% Copper, are intended to reduce erosion with copper bits but not so good with the iron plated bits.
'Not sure about the trend to high tin alloys with low or zero lead. I hear from associates that erosion has been worse, even hollowing, with iron plated tips.
Whatever the case, if you use an iron regularly, you'll be replacing bits likewise. I get up to a year but service guys get a lot less with daily use. 🙁
It depends on the bit quality. Cheap $10 hobby irons use plain copper rod with thin Nickel plating and erode slowly down to a wire from day 1. Seriously, you can file these to any shape and keep using them until nothing is left, if you wish.
Expensive bits, like those on temp. controlled irons, are usually thick plated with iron and can begin to erode suddenly when the core is exposed, depending on the solder alloy, flux and plating quality. Tin/lead Solders with ~ 0.5% Copper, are intended to reduce erosion with copper bits but not so good with the iron plated bits.
'Not sure about the trend to high tin alloys with low or zero lead. I hear from associates that erosion has been worse, even hollowing, with iron plated tips.
Whatever the case, if you use an iron regularly, you'll be replacing bits likewise. I get up to a year but service guys get a lot less with daily use. 🙁
No.
The classic filter formula only applies when two conditions are met. Rs = 0r0 and Rload = infinity.
Your example with 9k5 in the upper leg is loaded with Rload = 500r//220pF. That cannot be calculated by the classic formula.
You will find that adopting the actual output impedance of the Pot (treating it as the Source resistance) will give a closer approximation for the filtering effect.
Someone who is good at the maths will be able to give us the formula that takes account of the //500r. I can't.
Last edited:
Well, my soldering iron just bit the dust, so there's nothing more I can do for now.
- keantoken
- keantoken
This week's progress
The unused spaces are for phase lead/lag etc and zeners to protect circuit if trimmers should go open circuit. Going to fiddle to see what I prefer on the PCB version. Looks much better in real life - the flash just makes it look plain weird! Included Mike's rail filter too...
The unused spaces are for phase lead/lag etc and zeners to protect circuit if trimmers should go open circuit. Going to fiddle to see what I prefer on the PCB version. Looks much better in real life - the flash just makes it look plain weird! Included Mike's rail filter too...
Attachments
Last edited:
Well the pcb is a complete success, the only mistake was a reversed polarity on an electro....boy they go off with a bang!
I must say it still tickles me every time I hear one of these...
Only listened at low volume so far, but mike's filter is looking worthwhile...
I must say it still tickles me every time I hear one of these...
Only listened at low volume so far, but mike's filter is looking worthwhile...
Nice work Greg,
I look forward to hearing your impressions after some more listening.
All of my audio equipment & components have now arrived from the US so I can take a look at some square waves and make some meaningful comparisons to my other amp and with my DAC.
I'll begin populating my boards next week.
cheers
mike
I look forward to hearing your impressions after some more listening.
All of my audio equipment & components have now arrived from the US so I can take a look at some square waves and make some meaningful comparisons to my other amp and with my DAC.
I'll begin populating my boards next week.
cheers
mike
Great board, Greg, very compact and neat!
I have received word from my supplier that because I omitted Co Ltd on the net transfer the money cannot be paid into their account. I was never told; I asked about it constantly, only to be told this yesterday. I have set in train correction and I'm hoping the pcbs will be released next week. This has been a trial. The combined currency costs and bank fees come to 17% of the total sum. I swear I will never do business with anyone overseas again unless they have a Paypal account. PP is expensive, but it works instantly, and there are no quibbles about deleted 'Co., Ltd'.
Grimly Smiling,
Hugh
I have received word from my supplier that because I omitted Co Ltd on the net transfer the money cannot be paid into their account. I was never told; I asked about it constantly, only to be told this yesterday. I have set in train correction and I'm hoping the pcbs will be released next week. This has been a trial. The combined currency costs and bank fees come to 17% of the total sum. I swear I will never do business with anyone overseas again unless they have a Paypal account. PP is expensive, but it works instantly, and there are no quibbles about deleted 'Co., Ltd'.
Grimly Smiling,
Hugh
Thanks for the kind words. The photo really does not do it justice unfortunately. I think I may make it even smaller as I am leaning towards no compensation whatsoever, so those empty spaces are wasted. As it is the board is only 2.5" x 3.8".
Most disappointing to hear about your board issues Hugh, but it is good to know it seems to be sorted now. We're all eagerly awaiting them.
I'm looking forward to seeing what the masses think of the amplifier. Hopefully it will live up to expectations!
Good news Mike. I'm pretty sure even last night I could hear an improvement in noise levels - very little hiss from the speaker.
Kean, how long til you start work again?
Most disappointing to hear about your board issues Hugh, but it is good to know it seems to be sorted now. We're all eagerly awaiting them.
I'm looking forward to seeing what the masses think of the amplifier. Hopefully it will live up to expectations!
Good news Mike. I'm pretty sure even last night I could hear an improvement in noise levels - very little hiss from the speaker.
Kean, how long til you start work again?
I just ordered a new tip from Mouser, and it should be here in a few days, maybe even less since I'm near a Mouser plant.
My signal generator is determined to stay broken, without a schematic it is difficult to make any diagnosis. But I'm hoping it's bad electrolytics. Maybe a shorted tantalum somewhere. I replaced all the supply electros, which were too warm for comfort if not nearly shorted. It worked for a while with some quirks, but not since I heard a faint crackling noise in the corner... If only I could get it running I could make some more detailed tests of this amp. I have resorted to using the calibrator on my scope for square wave input.
My experimentation so far has led here:
1: VN0106 input had a cold and unpleasant sound. 2N5564 input brought more warmth and enjoyment, and subtlety.
2: Without miller or lead compensation, the amp sounds good, but from experience with my own amp it is possible to have cleaner treble. I am using an inferior VAS, which will be remedied when my order arrives. I ordered some VP0104, not sure what I was thinking, should have gone with a higher voltage version or the VP2110.
My iron first started eroding when I used hemostats to crush off the tip cleaner residue. I scored all the way around the tip in the process. I don't know much about soldering, besides point and shoot, but what I've been doing is using the tip cleaner/tinner can between soldering. I'm guessing that's wrong, and I should really find a sponge like everyone else.
On another topic, I'm experimenting with wax. I have a block of paraffin wax at my bench. I had this idea when I saw ancient Soshin micas covered in something like wax, from an old TV. My thought was perhaps it was to keep the moisture out, to keep from ruining the mica. I also thought it might help to protect from solder heat. If the components heats up, the wax will melt and evaporate, dissipating heat into the air. Does this make sense?
- keantoken
My signal generator is determined to stay broken, without a schematic it is difficult to make any diagnosis. But I'm hoping it's bad electrolytics. Maybe a shorted tantalum somewhere. I replaced all the supply electros, which were too warm for comfort if not nearly shorted. It worked for a while with some quirks, but not since I heard a faint crackling noise in the corner... If only I could get it running I could make some more detailed tests of this amp. I have resorted to using the calibrator on my scope for square wave input.
My experimentation so far has led here:
1: VN0106 input had a cold and unpleasant sound. 2N5564 input brought more warmth and enjoyment, and subtlety.
2: Without miller or lead compensation, the amp sounds good, but from experience with my own amp it is possible to have cleaner treble. I am using an inferior VAS, which will be remedied when my order arrives. I ordered some VP0104, not sure what I was thinking, should have gone with a higher voltage version or the VP2110.
My iron first started eroding when I used hemostats to crush off the tip cleaner residue. I scored all the way around the tip in the process. I don't know much about soldering, besides point and shoot, but what I've been doing is using the tip cleaner/tinner can between soldering. I'm guessing that's wrong, and I should really find a sponge like everyone else.
On another topic, I'm experimenting with wax. I have a block of paraffin wax at my bench. I had this idea when I saw ancient Soshin micas covered in something like wax, from an old TV. My thought was perhaps it was to keep the moisture out, to keep from ruining the mica. I also thought it might help to protect from solder heat. If the components heats up, the wax will melt and evaporate, dissipating heat into the air. Does this make sense?
- keantoken
Interesting I found something similar. I found the following:
1) Mosfet input + bjt vas = good
2) Mosfet input + mosfet vas = terrible
3) Jfet input + bjt vas = terrible
4) jfet input + mosfet vas = good
Why are you worried about damaging things while soldering? I have soldered hundreds of actives this last year and not lost a single one to soldering damage, and my technique is terrible!
1) Mosfet input + bjt vas = good
2) Mosfet input + mosfet vas = terrible
3) Jfet input + bjt vas = terrible
4) jfet input + mosfet vas = good
Why are you worried about damaging things while soldering? I have soldered hundreds of actives this last year and not lost a single one to soldering damage, and my technique is terrible!
I'm not worried about soldering, except perhaps when using polystyrene caps, which melt easily. Here it could be useful. Could be useful for sticking things to a ground plane too...
- keantoken
- keantoken
When you filter the frontend you essentially filter out the HF signal present on the rails. This signal the voltage drop produced across the PSU cap ESR by the output stage current. The fidelity of this signal is questionable since the ESR of these caps could be very nonlinear. Thus when you filter it out you may be removing it's own set of intermodulation signals. Intermodulation may equate with noise in this case. Does this sound like an apt explanation?
- keantoken
- keantoken
I'm hoping that when you listen to music at normal levels that it you will also notice the overall sound is more musical and natural.
Much of the noise you have reduced is way beyond the audible range but the effects of it somehow negatively impact the audio output in the form of background grunge.
I read somewhere that amps have an uncanny knack of being able to demodulate ultrasonic noise into the audio spectrum.
Hello KT
Which brand and model number are your signal generator (I may be lucky and find you the schematic) ?
Bye
Gaetan
- Status
- Not open for further replies.