No question, no cure, no end
Considering ways to keep a component flexible after 'completion' is worse than fanatical -- it is obsessive!
R-cores arrived!
The r-cores arrived and appear to be perfectly suited. I will begin to test them under load tomorrow and measure voltage drops across the various components. The tiny gold banana plug socket pairs arrived as well and are just what I had in mind for allowing the regulator modules to be changed easily. My goal tomorrow will be to assure that the entire power train comes up to voltage under load without causing any of the components or their heatsinks to overheat. This ignores the temperature of the MOSFET / Caddock sinks but we already know they run hot.
Several components for the JC-80 remain in-transit but when they arrive everything will finally be in-house. Wolfsin now knows a source for every component assuming the r-cores will hold the required 30vDC before the cap multipliers.
The r-cores arrived and appear to be perfectly suited. I will begin to test them under load tomorrow and measure voltage drops across the various components. The tiny gold banana plug socket pairs arrived as well and are just what I had in mind for allowing the regulator modules to be changed easily. My goal tomorrow will be to assure that the entire power train comes up to voltage under load without causing any of the components or their heatsinks to overheat. This ignores the temperature of the MOSFET / Caddock sinks but we already know they run hot.
Several components for the JC-80 remain in-transit but when they arrive everything will finally be in-house. Wolfsin now knows a source for every component assuming the r-cores will hold the required 30vDC before the cap multipliers.
As might be surmised from the delay, the r-cores are overvoltage. I looked at chokes to reduce the voltage but one's choices come down to 90% (instead of 141% using caps) for large chokes or somewhere in between with lots of noise. Resistors can be substituted for thermistors readily, are cheap, and show less noise in a simulation.
Chokes have lower MAX-MIN range
Even in the steady state there appears to be a lot of jitter in the choke case but the difference in ripple voltage is less overall (by about 10%) using a 25mH choke than a 10r resistor right after the rectifiers. In addition, the effect of a common mode choke at filtering noise across both rails is not being simulated. What I observed may be a function of the simulator itself, as it takes much longer to run, leading me to believe it is using much smaller time intervals.
Even in the steady state there appears to be a lot of jitter in the choke case but the difference in ripple voltage is less overall (by about 10%) using a 25mH choke than a 10r resistor right after the rectifiers. In addition, the effect of a common mode choke at filtering noise across both rails is not being simulated. What I observed may be a function of the simulator itself, as it takes much longer to run, leading me to believe it is using much smaller time intervals.
I was able to spice the jims audio circuit quite successfully to understand the bias/dissipation and how to adjust the gain.
BUT! I know that the models I have for the semiconductors are poor at best and I should treat any results from the simulation accordingly. If anyone has models they trust for these devices it would be good to share them.
On another matter altogether - does anyone know a source of heatsinks for the cap multiplier mosfets? The pin spacing on the board seems a bit unusual.....
BUT! I know that the models I have for the semiconductors are poor at best and I should treat any results from the simulation accordingly. If anyone has models they trust for these devices it would be good to share them.
On another matter altogether - does anyone know a source of heatsinks for the cap multiplier mosfets? The pin spacing on the board seems a bit unusual.....
Good Work!
Would you be willing to post your spice file? Which component models do you mistrust?
>>On another matter altogether - does anyone know a source of heatsinks for the cap multiplier mosfets?
I have not yet but am still looking. For now I am tapping one hole and using a single 2-56 screw to anchor it. Please post source if you find it.
Would you be willing to post your spice file? Which component models do you mistrust?
>>On another matter altogether - does anyone know a source of heatsinks for the cap multiplier mosfets?
I have not yet but am still looking. For now I am tapping one hole and using a single 2-56 screw to anchor it. Please post source if you find it.
I've used the closest pin spacing for this type of heatsink but I grinded them from the outside in afterward. Left a half of the pin thickness and solder! Worked fine! I have some small 1" and 1.25" chinese heatsinks that are made to fit exactly on those holes but are too small for the dissipated heat... Most of Jim's Audio boards use these pin spacing, which are not standard anywhere else in the world it seems...
Supply Simulation
First note that you will need dual rails at + and - 30 vDC. The preamp bias current is adjustable and is usually set in the 200ma to 300ma range on each rail. Each JC-80 will require as much as 18 watts but with various losses your transformer will need to be at least 25va per channel. I start by mentioning this because sourcing appropriately sized r-core transformers may become challenging and, if the item(s) you receive are undersized and non-returnable, expensive. Transformer(s) comprise roughly a fourth of the cost of the entire project and their characteristics impact some other components.
Many configurations of transformers are supported by the rectifier pcb, including dual secondary, and center tapped. The transformer I settled on was advertised as dual 30VAC secondaries @.5A. In the no load case it measured 37VAC. The resistance of each secondary is 6 ohms. With no more than this data you can use PSU Designer II to simulate the rectifier and capacitor board behavior. I have attached the resistor design, with ten ohm resistors, and the 25mH choke design. I have included higher voltage diodes with the choke because the simulator detects high spikes from it. The voltage across the capacitors must be enough above 30VDC to allow for regulator drop so both of those designs provide roughly nine volts over. Keep in mind that we are simulating only one rail so, especially with center tapped transformers, YMMV.
My recommendation would be that you google PSU Designer and download the simulator so you can understand the effects of changes in component values. Then select candidate transformers and when they arrive simulate again to determine with greater precision the values of resistors or common mode chokes you need for the voltage you desire.
First note that you will need dual rails at + and - 30 vDC. The preamp bias current is adjustable and is usually set in the 200ma to 300ma range on each rail. Each JC-80 will require as much as 18 watts but with various losses your transformer will need to be at least 25va per channel. I start by mentioning this because sourcing appropriately sized r-core transformers may become challenging and, if the item(s) you receive are undersized and non-returnable, expensive. Transformer(s) comprise roughly a fourth of the cost of the entire project and their characteristics impact some other components.
Many configurations of transformers are supported by the rectifier pcb, including dual secondary, and center tapped. The transformer I settled on was advertised as dual 30VAC secondaries @.5A. In the no load case it measured 37VAC. The resistance of each secondary is 6 ohms. With no more than this data you can use PSU Designer II to simulate the rectifier and capacitor board behavior. I have attached the resistor design, with ten ohm resistors, and the 25mH choke design. I have included higher voltage diodes with the choke because the simulator detects high spikes from it. The voltage across the capacitors must be enough above 30VDC to allow for regulator drop so both of those designs provide roughly nine volts over. Keep in mind that we are simulating only one rail so, especially with center tapped transformers, YMMV.
My recommendation would be that you google PSU Designer and download the simulator so you can understand the effects of changes in component values. Then select candidate transformers and when they arrive simulate again to determine with greater precision the values of resistors or common mode chokes you need for the voltage you desire.
Squeezin'
While both resistors and chokes are in transit, Wolfsin looked carefully at the rectifier pcb. He noted that 10x2200uf 'lytics would fit easily, 10x3300uf could be squeezed in, but by careful staggering of 2200 & 6800uf caps it would be possible to place 40,400uf on that same pcb. The simulator shows doing that on a channel with chokes rather than resistors cuts the ripple in HALF not even considering the common mode benefits.
I know ticknpop was planning to simulate the tracking prereg
and Do's silent partner is doing same for JC-80, so we might squeeze some juice yet!
While both resistors and chokes are in transit, Wolfsin looked carefully at the rectifier pcb. He noted that 10x2200uf 'lytics would fit easily, 10x3300uf could be squeezed in, but by careful staggering of 2200 & 6800uf caps it would be possible to place 40,400uf on that same pcb. The simulator shows doing that on a channel with chokes rather than resistors cuts the ripple in HALF not even considering the common mode benefits.
I know ticknpop was planning to simulate the tracking prereg
and Do's silent partner is doing same for JC-80, so we might squeeze some juice yet!Kubota dropout measured
The 33mH common mode choke arrived and I built the loading jig today. The jig provides a dummy load after the regulator (one 10 watt 150r resistor from each rail to ground). At 30vDC this allows 200ma to flow. Without load the voltage across the caps soars. Even with 1k bleeder resistors my caps reach 45v so hot plugging a loaded regulator might not be pretty.
The dropout for the Kubota is 4.9-5v. I have not yet assembled a tracking prereg but already know to expect a dropout >6vDC. While I want to keep as much dissipated heat as possible in the external enclosure, I also want to be able to swap regulators easily without altering the inductance. That means I want the loaded voltage at the output of the rectifier and caps board to be at least 30v + 6.5v (the dropout of the regulator with the larger dropout). Leaving a little extra will allow for error at a cost of a little more heat dissipation from the regulator in the same case as the preamp.
The 33mH common mode choke arrived and I built the loading jig today. The jig provides a dummy load after the regulator (one 10 watt 150r resistor from each rail to ground). At 30vDC this allows 200ma to flow. Without load the voltage across the caps soars. Even with 1k bleeder resistors my caps reach 45v so hot plugging a loaded regulator might not be pretty.
The dropout for the Kubota is 4.9-5v. I have not yet assembled a tracking prereg but already know to expect a dropout >6vDC. While I want to keep as much dissipated heat as possible in the external enclosure, I also want to be able to swap regulators easily without altering the inductance. That means I want the loaded voltage at the output of the rectifier and caps board to be at least 30v + 6.5v (the dropout of the regulator with the larger dropout). Leaving a little extra will allow for error at a cost of a little more heat dissipation from the regulator in the same case as the preamp.
My Components
1 r-core 2x37vAC secondary
8 FFP20UP20DN Soft recovery rectifiers
2 thermistors
1 39mH CM choke your inductance TBD
6 2200ufd 50v
4 6800ufd 50v
2 1k 5w bleeder resistor
1 Dual Rail Linear Power Supply PCB
1 fuse 250ma
Notes:
1) Regulator out voltage depends on regulator dropout; select transformer first then similate TBD's. if 30v secondaries, CM choke not possible
2) TBDs include choke inductance, capacitor voltage
3) pcb comes with add'l flex & written docs
4) fuse each channel separately
5) ebay advanced search 'by seller' audiosector for pcb and ssrecord for NOS talema common mode chokes
1 r-core 2x37vAC secondary
8 FFP20UP20DN Soft recovery rectifiers
2 thermistors
1 39mH CM choke your inductance TBD
6 2200ufd 50v
4 6800ufd 50v
2 1k 5w bleeder resistor
1 Dual Rail Linear Power Supply PCB
1 fuse 250ma
Notes:
1) Regulator out voltage depends on regulator dropout; select transformer first then similate TBD's. if 30v secondaries, CM choke not possible
2) TBDs include choke inductance, capacitor voltage
3) pcb comes with add'l flex & written docs
4) fuse each channel separately
5) ebay advanced search 'by seller' audiosector for pcb and ssrecord for NOS talema common mode chokes
If you are talking about the Kubota regulator, no. In the case of the JC-80, a BOMless partial is toward the end of *Èý×êÐÅÓþ*Dennesen JC-80ȫƽºâÇ°¼¶¶Æ½ðPCB°åÿÉùµÀ-ÌÔ±¦Íø. The schematic of the rectifier pcb is distributed in hard copy with that board. Which one interests you?
Oh, sorry about that. The JC-80 schematic is what I was looking for.
I have schematics and boards for a couple of different versions of the Kubota regulator.
I also have a couple of Borbely super shunts in use. It seems that regardless of what I monkey with they are the defacto reference for uses under 200ma.
The JC-80 looks like a jfet complementary, differential to mosfet source follower with dc servo??? Hard to tell.
Certainly there are complementary jfets, it is differential, the outputs are mosfets, and it employs servo circuitry. John Curl calls it a transconductance amplifier. There are two contributors who've completed and are listening and appreciating thier units but since neither has responded to your question I am doing my best.
For the schematics, it is another story... It is a John Curl IP and the only way you can get them is if you ask John or buy the PCBs from the Ebay resellers.
We all have the Chinese PCBs and schematics that are distributed with them but without author approval I do not feel comfortable handing them out...
Hope you understand
Thanks
Do
We all have the Chinese PCBs and schematics that are distributed with them but without author approval I do not feel comfortable handing them out...
Hope you understand
Thanks
Do
Thanks Do. I think hags got what he needed from the schematic posted on the web but maybe not. I agree we should not divulge more than is already 'out there'.
The eBay cm choke vendor agreed to a matched pair of 39mH cm chokes and Wolfsin izAbout to test and measure incircuit, under load. I finished the long cut required along the edge of the main chassis mount. As soon as I invent a reliable banana holder I'll look for a peel to slip on!
The eBay cm choke vendor agreed to a matched pair of 39mH cm chokes and Wolfsin izAbout to test and measure incircuit, under load. I finished the long cut required along the edge of the main chassis mount. As soon as I invent a reliable banana holder I'll look for a peel to slip on!
JC-80 gain
Here's the latest news from my friend's PSpice/Orcad simulation...
Setting the gain
"If you reduce the value of R17, R18, R19 and R20 to 330 ohms,
it will drop the gain to about 5.7x (15.2dB). This would
cause the overall bias to increase somewhat, so you'll need
to reduce the R41 trimpot adjustment a bit to get back to
230mA current draw."
Using IRF610/IRF9610
"Running at the lower gain also increases the frequency response
to -3dB at almost 15MHz. This is in part also due to the
IRF610/9610's lower gate capacitance compared to IRF510/9510.
Anyway, using the IRF610/9610 does not otherwise change the
operating points or the gain of the amp."
Asking if in simulation, IRF610/9610 was worse than IRF510/9510
"No, IRF610/9610 is not worse than IRF510/9510. It has lower gate
capacitance so the bandwidth is wider. Reducing the gain also
makes the bandwidth wider because an amp usually has a fairly
constant gain-bandwidth product, that is, as you decrease the
gain, the bandwidth naturally increases.
Having a wide-band amp is good, 2MHz is superb, but 15MHz is
not 7.5x better. We could only hear signals to 20KHz (if
that much), but usually 10x to 100x that frequency is more than
enough to ensure optimum performance within the audio band,
such as minimum phase shift and distortion, etc. Too much
bandwidth may make the amp act as a radio receiver in certain
circumstances.
Decreasing the gain should not degrade the sound. If anything,
it should lower the noise floor significantly."
The 2MHz was when running the simulation with 15.3x gain and IRF510/IRF9510...
I will follow up with some graphs shortly!
Thanks
Do
Here's the latest news from my friend's PSpice/Orcad simulation...
Setting the gain
"If you reduce the value of R17, R18, R19 and R20 to 330 ohms,
it will drop the gain to about 5.7x (15.2dB). This would
cause the overall bias to increase somewhat, so you'll need
to reduce the R41 trimpot adjustment a bit to get back to
230mA current draw."
Using IRF610/IRF9610
"Running at the lower gain also increases the frequency response
to -3dB at almost 15MHz. This is in part also due to the
IRF610/9610's lower gate capacitance compared to IRF510/9510.
Anyway, using the IRF610/9610 does not otherwise change the
operating points or the gain of the amp."
Asking if in simulation, IRF610/9610 was worse than IRF510/9510
"No, IRF610/9610 is not worse than IRF510/9510. It has lower gate
capacitance so the bandwidth is wider. Reducing the gain also
makes the bandwidth wider because an amp usually has a fairly
constant gain-bandwidth product, that is, as you decrease the
gain, the bandwidth naturally increases.
Having a wide-band amp is good, 2MHz is superb, but 15MHz is
not 7.5x better.
We could only hear signals to 20KHz (ifthat much), but usually 10x to 100x that frequency is more than
enough to ensure optimum performance within the audio band,
such as minimum phase shift and distortion, etc. Too much
bandwidth may make the amp act as a radio receiver in certain
circumstances.
Decreasing the gain should not degrade the sound. If anything,
it should lower the noise floor significantly."
The 2MHz was when running the simulation with 15.3x gain and IRF510/IRF9510...
I will follow up with some graphs shortly!
Thanks
Do
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.