The idea to stick in the oil cap as the first transitional cap came from some posts I read from Dave Slagle.
That worked out well, so I went looking for other Slagle power supply ideas. On Dave's forum I read a post where he suggested to a DIYer to remove his final LC cell. And on his index page, he says "We have had great results with CLCLC power supplies with the first choke being Amorphous and the second choke being Nickel."
So even though I had to move around some alligator clips, I decided to make the two minute herculean effort to pull out my last LC stage and make it a CLCLC.
This turned out to be a good win. There were noticeable gains in tonality, frequency extension, and dynamics. Obvious gains, not requiring any back and forth checking at all. Definitely a bandwidth bottleneck removed.
Plus the power supply is smaller and cheaper by removing a big capacitor and a small inductor. I did end up using the 70 uF Oil Obbligato Oil/Film as the last cap instead of the 47 uF Obbligato Film as I thought it sounded better in my circuit.
So the new CLCLC supply with the current Soviet Special Aikido is "Configuration D". As usual, I'm following the Swenson methodology for getting a good impulse response.
Ripple is higher, of course. PSUD simulated ripple is 20+ mV. Interestingly, actual measured ripple is only 12 mV and steady on the scope. This is higher than the previous supply's 3-4 mV but sounds subjectively ok to me on horns.
The only downside, other than a touch more hum, is a bit of sizzle in the treble but I believe this is the class-d amp showing it's true colors. I'll revert back to my 300b amp soon.
Given that Dave's first two suggestions worked well, I'm guessing using an Amorphous core first inductor and Nickel core second inductor would be good too, but that is $$$ and must wait.
When I get my courage up I'll try running some FFTs.
Here is the new supply showing the 10% change in current. Note impulse response is quicker than the previous supply.
That worked out well, so I went looking for other Slagle power supply ideas. On Dave's forum I read a post where he suggested to a DIYer to remove his final LC cell. And on his index page, he says "We have had great results with CLCLC power supplies with the first choke being Amorphous and the second choke being Nickel."
So even though I had to move around some alligator clips, I decided to make the two minute herculean effort to pull out my last LC stage and make it a CLCLC.
This turned out to be a good win. There were noticeable gains in tonality, frequency extension, and dynamics. Obvious gains, not requiring any back and forth checking at all. Definitely a bandwidth bottleneck removed.
Plus the power supply is smaller and cheaper by removing a big capacitor and a small inductor. I did end up using the 70 uF Oil Obbligato Oil/Film as the last cap instead of the 47 uF Obbligato Film as I thought it sounded better in my circuit.
So the new CLCLC supply with the current Soviet Special Aikido is "Configuration D". As usual, I'm following the Swenson methodology for getting a good impulse response.
Ripple is higher, of course. PSUD simulated ripple is 20+ mV. Interestingly, actual measured ripple is only 12 mV and steady on the scope. This is higher than the previous supply's 3-4 mV but sounds subjectively ok to me on horns.
The only downside, other than a touch more hum, is a bit of sizzle in the treble but I believe this is the class-d amp showing it's true colors. I'll revert back to my 300b amp soon.
Given that Dave's first two suggestions worked well, I'm guessing using an Amorphous core first inductor and Nickel core second inductor would be good too, but that is $$$ and must wait.
When I get my courage up I'll try running some FFTs.
Here is the new supply showing the 10% change in current. Note impulse response is quicker than the previous supply.
Attachments
Hi Cappy
I was hoping you could answer a slightly off topic question for me.
I am planning to build a TP DAC, but am trying to decide which one. Wondering if I should build a Buffalo, COD or Opus, and I noticed you had replaced the COD with the Buffalo.
How would you compare the two?
I was leaning towards a Opus, in dual mono config, with two Ivy's, although I may wait for the in development discrete IV module. Source is a squeezebox into the metronome.
TIA
Randy
I was hoping you could answer a slightly off topic question for me.
I am planning to build a TP DAC, but am trying to decide which one. Wondering if I should build a Buffalo, COD or Opus, and I noticed you had replaced the COD with the Buffalo.
How would you compare the two?
I was leaning towards a Opus, in dual mono config, with two Ivy's, although I may wait for the in development discrete IV module. Source is a squeezebox into the metronome.
TIA
Randy
Randy,
I prefer the Buffalo, at least with SPDIF.
I did about 20 minutes of listening with the COD via the I2S of the Pace Car and it was very good. I had a grounding problem -- too many different boxes I think -- and there was some hum but I didn't get around to debugging that. Now the COD has swum up the Willamette River to Portland - a friend is enjoying it with a headphone setup.
In SPDIF mode I just like the resolution and accurate tone of the Buffalo. The COD sounded a bit fuzzier.
Your post got me to reinstall the Buffalo with Configuration "D" tonight. Nickel Creek is sounding quite fine as I write this. I can say with some relief the bass is good too -- it sounded a bit "porous" previously. This is with passive I/V of 15 ohms through one half of the differential output. Nothing at all changed with the Buffalo, just the 6n6p swap and the major power supply changes. I'm going to have to do some measurements with the Buffalo with the new and old power supply to see if I can figure out what was going on.
I prefer the Buffalo, at least with SPDIF.
I did about 20 minutes of listening with the COD via the I2S of the Pace Car and it was very good. I had a grounding problem -- too many different boxes I think -- and there was some hum but I didn't get around to debugging that. Now the COD has swum up the Willamette River to Portland - a friend is enjoying it with a headphone setup.
In SPDIF mode I just like the resolution and accurate tone of the Buffalo. The COD sounded a bit fuzzier.
Your post got me to reinstall the Buffalo with Configuration "D" tonight. Nickel Creek is sounding quite fine as I write this. I can say with some relief the bass is good too -- it sounded a bit "porous" previously. This is with passive I/V of 15 ohms through one half of the differential output. Nothing at all changed with the Buffalo, just the 6n6p swap and the major power supply changes. I'm going to have to do some measurements with the Buffalo with the new and old power supply to see if I can figure out what was going on.
The FFT plots with no signal, was the input shorted or left open?
I also use a 192 with AudioTester, and can't seem to get the zero dB level where I want it. It's excellent software but has some odd blind spots. The card input should be buffered (I use a BUF03 for that), and for high level measurement (>2V), you can use a 10x scope probe to knock down the levels. Set the buffer input resistor to 1M and whatever your scope's input capacitance is.
I also use a 192 with AudioTester, and can't seem to get the zero dB level where I want it. It's excellent software but has some odd blind spots. The card input should be buffered (I use a BUF03 for that), and for high level measurement (>2V), you can use a 10x scope probe to knock down the levels. Set the buffer input resistor to 1M and whatever your scope's input capacitance is.
Sy,
They were open. Should I short them for the next batch of measurements?
Regarding the zero dB level. Is it perhaps because you can't run the calibration routine? I can't get that to work. I need to email the developer on that possible bug. It's a pretty big issue, methinks.
Good idea on the scope probe, thanks.
Regarding buffering of the the card input. I will do that. What will that accomplish? Is it for safety or will it also change the measurements because it normalizes impedance?
Regards,
Bill
They were open. Should I short them for the next batch of measurements?
Regarding the zero dB level. Is it perhaps because you can't run the calibration routine? I can't get that to work. I need to email the developer on that possible bug. It's a pretty big issue, methinks.
Good idea on the scope probe, thanks.
Regarding buffering of the the card input. I will do that. What will that accomplish? Is it for safety or will it also change the measurements because it normalizes impedance?
Regards,
Bill
Buffalo - Aikido Interface
Hi Cappy and others,
I'd like to hear the details in how you interfaced the Buffalo's output to the Aikido. Cappy it seems like you are running singled ended, did you use both (+/-) DAC outputs for each channel?
I've been considering either going the balanced Aikido route or implementing a Broskie Cathode Follower.
Thanks.
-Eric
Hi Cappy and others,
I'd like to hear the details in how you interfaced the Buffalo's output to the Aikido. Cappy it seems like you are running singled ended, did you use both (+/-) DAC outputs for each channel?
I've been considering either going the balanced Aikido route or implementing a Broskie Cathode Follower.
Thanks.
-Eric
Eric,
Right, I'm running single ended.
So far I've just pulled the signal from the + differential output.
From the + and ground outputs, the dac output goes directly into the grid of the first Aikido stage. I did not find a post-dac analog filter necessary with the Aikido (other than the implicit one), which is nice.
The balanced Aikido would be cool if one has a balanced preamp/amp.
Looking at the schematic for the Broskie Cathode Follower, if I understand it correctly, it looks like a circuit for voltage out DACs. I believe the Sabre supports voltage out, but does the Buffalo?
Right, I'm running single ended.
So far I've just pulled the signal from the + differential output.
From the + and ground outputs, the dac output goes directly into the grid of the first Aikido stage. I did not find a post-dac analog filter necessary with the Aikido (other than the implicit one), which is nice.
The balanced Aikido would be cool if one has a balanced preamp/amp.
Looking at the schematic for the Broskie Cathode Follower, if I understand it correctly, it looks like a circuit for voltage out DACs. I believe the Sabre supports voltage out, but does the Buffalo?
I believe I read earlier that you are using a 15 ohm I/V resistor? Does the Buffalo develop sufficient voltage to drive the grid directly?
I'm going to be building a balanced input amp so the Balanced Aikido will be the ticket.
In the interim I'm thinking of going with a two stage Balanced-to-SE converter. Each of the +/- DAC outputs would feed a separate grounded grid stage (low input impedance). The output of these stages would be combined in a Broskie Cathode follower (as JB shows for voltage DACS). This should be straight-forward to implement.
I definitely plan on leveraging your PSU work Cappy, thanks for sharing.
Right, a 15 ohm resistor. And actually that is too much gain for my system! Five ohms would probably be about right to get my volume control in the 12 o'clock to 4 o'clock TVC sweet spot. The output of the Buffalo is strong.
Yeah, with a balanced amp the balanced Aikido sounds right. You probably know the heaters will be pulling a lot of current, up to 6 amps, depending on tube choice. It should look impressive. I definitely am going to have my case have the tubes on the outside. For you, it might be a requirement unless you do a mesh type case.
Ok, I see what you are planning for the Broskie Cathode Follower.
You are welcome. It has been very useful for me to write down my thoughts as I think it has focused my efforts. I think you will be pleased with a similar power supply.
Yeah, with a balanced amp the balanced Aikido sounds right. You probably know the heaters will be pulling a lot of current, up to 6 amps, depending on tube choice. It should look impressive. I definitely am going to have my case have the tubes on the outside. For you, it might be a requirement unless you do a mesh type case.
Ok, I see what you are planning for the Broskie Cathode Follower.
You are welcome. It has been very useful for me to write down my thoughts as I think it has focused my efforts. I think you will be pleased with a similar power supply.
Previously I mentioned I wasn't able to use 6CG7s and 5687s with the Tentlabs IDHT supplies as they couldn't give 6.3 vdc out -- the current was too high for my input ac voltage. I was able to pull 1 amp each on the supplies so was using 6n23p-evs in the first postion, since they have a smaller heater requirement.
A few days ago I installed another filament transformer with higher voltage. I'm still not able to get the supplies to put out 6.3vdc, however. They are spec'd up to 1.5 amps draw.
I wanted to mention this as a heads-up in case anyone else is thinking of using the supplies with the Aikido. I have an email in to Guido -- hopefully he will be able to figure out why this isn't working in my situation.
A few days ago I installed another filament transformer with higher voltage. I'm still not able to get the supplies to put out 6.3vdc, however. They are spec'd up to 1.5 amps draw.
I wanted to mention this as a heads-up in case anyone else is thinking of using the supplies with the Aikido. I have an email in to Guido -- hopefully he will be able to figure out why this isn't working in my situation.
Cappy,
Another small diversion and perhaps a stupid question, but is it possible to build the aikidos for lower gain such as 0 dB, 3dB or 6dB to be used exclusively as a buffer amp, or is the gain structure fixed in the topology?
The answer is probably somewhere on JB's site and either I can't find it, or am reading past it.
-- Jim
Another small diversion and perhaps a stupid question, but is it possible to build the aikidos for lower gain such as 0 dB, 3dB or 6dB to be used exclusively as a buffer amp, or is the gain structure fixed in the topology?
The answer is probably somewhere on JB's site and either I can't find it, or am reading past it.
-- Jim
Jim,
The first tube position fixes the gain. Well not exactly, but the second tube has little impact. The total gain works out to be about half the mu of the first tube.
Broskie gives the example of a 6CG7 followed by 6DJ8. With a B+ of 300v, and cathode resistors of 1k, a 6CG7 will have a gain of 10.1, while 6DJ8 a gain of 0.96, for total gain of 9.7, or just under 20 db.
Tubes with smaller "mu"s are 6n6p, etc. I don't know how well they would work in the first position and gain would still be 17+ db. So basically, the Aikido is set up as a voltage gain and buffer amp. If you are looking for a current buffer alone or current buffer with minimal gain there are probably less complex and better sounding circuits. I'd probably try a DHT.
The first tube position fixes the gain. Well not exactly, but the second tube has little impact. The total gain works out to be about half the mu of the first tube.
Broskie gives the example of a 6CG7 followed by 6DJ8. With a B+ of 300v, and cathode resistors of 1k, a 6CG7 will have a gain of 10.1, while 6DJ8 a gain of 0.96, for total gain of 9.7, or just under 20 db.
Tubes with smaller "mu"s are 6n6p, etc. I don't know how well they would work in the first position and gain would still be 17+ db. So basically, the Aikido is set up as a voltage gain and buffer amp. If you are looking for a current buffer alone or current buffer with minimal gain there are probably less complex and better sounding circuits. I'd probably try a DHT.
There is something about the 6n6p based Aikido that was starting to bug me. It's a "fuzziness", particularly in the bass.
I went back and forth between the 5687 and 6n6p Aikidos 4-5 times and I could hear the fuzziness pretty clearly.
Apart from that, I like the big 6n6p sound and nice midrange. But the 5687 is clearer and also has a nearly as nice midrange. I'm wondering if the big soundstage is due to phasey fuzziness.
I tried upping the current through the tube, trying Configuration E at 17.5 ma and Configuration F at 19 ma. That did help a bit, but not enough for my taste.
Perhaps using solid state rectification would help clear things up. This is the second circuit I've tried the 6n6p in, and fuzzy bass was a problem in the other circuit too. Also, it seems like people I know who have had success with the tube are using push-pull or differential configurations, not single-ended.
The 6n6p is voted off the island.
I went back and forth between the 5687 and 6n6p Aikidos 4-5 times and I could hear the fuzziness pretty clearly.
Apart from that, I like the big 6n6p sound and nice midrange. But the 5687 is clearer and also has a nearly as nice midrange. I'm wondering if the big soundstage is due to phasey fuzziness.
I tried upping the current through the tube, trying Configuration E at 17.5 ma and Configuration F at 19 ma. That did help a bit, but not enough for my taste.
Perhaps using solid state rectification would help clear things up. This is the second circuit I've tried the 6n6p in, and fuzzy bass was a problem in the other circuit too. Also, it seems like people I know who have had success with the tube are using push-pull or differential configurations, not single-ended.
The 6n6p is voted off the island.
I did more tube and resistor rolling:
Configuration F: 6n23p/6n23p
Too much of that straight up 6DJ8 sound. Thin.
Configuration G: 6n1p/6n23p
Ah.... no.
Configuration H: 6cg7/6n23p
Now this sounds better. I tried three different tubes: RCA cleartop, RCA blackplate, Mazda/Brimar. I like the Mazda best but still not balanced sounding enough for my tastes. All were a bit bland, too.
Configuration I: 6cg7/6n6p
I was able to try this with the Mazda tubes because they draw a lot less heater current than the other 6cg7s. Bland and fuzzy.
Configuration J: 6cg7/5687
I was able to try this with the Mazda 6cg7s also. The other 6cg7s cause the Tentlabs supplies to not maintain voltage. This sounds good.
I still prefer the 6n23p/5687 combination at this point though. It is more lively, better soundstage depth, more dynamic. Also, the gain of 6DJ8 variants is more copacetic with a DAC. The 6CG7 doesn't really have enough gain -- otherwise the I/V resistor needs to be bigger, which isn't optimal. I also noticed more noise with all three of my 6cg7s - and I don't think it is the tubes themselves generating the noise - I'm wondering if the grid conditions of the 6DJ8 favor use in a DAC more.
So the 6n23p-ev/5687 is most likely the final choice. I am listening to Configuration K now which is 7.5 mA on the 6n23ps and 18.5 mA on the 5687s. Just a touch mellower, which matches the nice aspects of the 6n6p better.
Configuration F: 6n23p/6n23p
Too much of that straight up 6DJ8 sound. Thin.
Configuration G: 6n1p/6n23p
Ah.... no.
Configuration H: 6cg7/6n23p
Now this sounds better. I tried three different tubes: RCA cleartop, RCA blackplate, Mazda/Brimar. I like the Mazda best but still not balanced sounding enough for my tastes. All were a bit bland, too.
Configuration I: 6cg7/6n6p
I was able to try this with the Mazda tubes because they draw a lot less heater current than the other 6cg7s. Bland and fuzzy.
Configuration J: 6cg7/5687
I was able to try this with the Mazda 6cg7s also. The other 6cg7s cause the Tentlabs supplies to not maintain voltage. This sounds good.
I still prefer the 6n23p/5687 combination at this point though. It is more lively, better soundstage depth, more dynamic. Also, the gain of 6DJ8 variants is more copacetic with a DAC. The 6CG7 doesn't really have enough gain -- otherwise the I/V resistor needs to be bigger, which isn't optimal. I also noticed more noise with all three of my 6cg7s - and I don't think it is the tubes themselves generating the noise - I'm wondering if the grid conditions of the 6DJ8 favor use in a DAC more.
So the 6n23p-ev/5687 is most likely the final choice. I am listening to Configuration K now which is 7.5 mA on the 6n23ps and 18.5 mA on the 5687s. Just a touch mellower, which matches the nice aspects of the 6n6p better.
Caught the Squirrel Nut Zippers live last night at the local and historic WOW (Woodmen of the World) Hall. I missed seeing them in the 90s and thought I had missed my chance. They were terrific! If they are touring through your town I highly recommend seeing them live.
The lower current Aikido version didn't last long. It does appear the 6DJ8 tubes like higher current flow, as designers like Allen Wright have mentioned.
I've been on Configuration L for a while, which is 9.3 mA on the 6n23p and 20.5 mA on the 5687. The cathode resistor on the 5687 is a 100 ohm Shinkoh and the 6n23p has a 150 ohm combo of Riken and Shinkoh. The B+ voltage is about 206 volts.
I've read a lot about resistors over the years on the various audio forums but I haven't run across anyone else mixing resistors on one circuit element to fine tune the sound. I'm sure others are doing it, though. I've also done it with success in my phono stage. There are few rules for adding spices to food or audio gear. I think it can make good sense for spices like sage and also for resistors which can have strong personalities.
The lower current Aikido version didn't last long. It does appear the 6DJ8 tubes like higher current flow, as designers like Allen Wright have mentioned.
I've been on Configuration L for a while, which is 9.3 mA on the 6n23p and 20.5 mA on the 5687. The cathode resistor on the 5687 is a 100 ohm Shinkoh and the 6n23p has a 150 ohm combo of Riken and Shinkoh. The B+ voltage is about 206 volts.
I've read a lot about resistors over the years on the various audio forums but I haven't run across anyone else mixing resistors on one circuit element to fine tune the sound. I'm sure others are doing it, though. I've also done it with success in my phono stage. There are few rules for adding spices to food or audio gear. I think it can make good sense for spices like sage and also for resistors which can have strong personalities.
Cappy said:
So what do the different resistors sound like? Riken versus Shinkoh versus others you use? Or, is it too complicated, because it depends where you put them?
I am modding a couple of 800b monoblocks that I just got, and need to buy some resistors anyway. I was going to use mostly Rikens, but had dales, mills, and Kiwames in the mix too.
Randy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.