Change the input voltage and for each new input voltage measure the CCS resistor Vdrop.
In a correctly operating CCS that resistor Vdrop should change by very little over a quite wide range of input voltage.
Then ensure that your input voltage never drops below the lowest that results in that nominal Vdrop.
This was the reason for the 7V recommendation in the earlier versions. Vin -Vout>=7V for proper CCS resistor Vdrop over a range of mains voltage variations.
Once you have adequate Vin, then you need to measure the Shunt current to ensure it never falls to near zero, when the load draws it's maximum demand. This is a bit more difficult. There are no resistors in the output, nor in the Shunt routes that allow comparison of currents in max demand operation.
You can monitor output voltage to see if there is a glitch (voltage droop) if the output demand increases. But this ONLY works if the CCS is working.
If the CCS is not working then a small voltage drop at the output increases the Vdrop across the CCS resistor and the CCS resistor is now effectively regulating for the output demand.
In a correctly operating CCS that resistor Vdrop should change by very little over a quite wide range of input voltage.
Then ensure that your input voltage never drops below the lowest that results in that nominal Vdrop.
This was the reason for the 7V recommendation in the earlier versions. Vin -Vout>=7V for proper CCS resistor Vdrop over a range of mains voltage variations.
Once you have adequate Vin, then you need to measure the Shunt current to ensure it never falls to near zero, when the load draws it's maximum demand. This is a bit more difficult. There are no resistors in the output, nor in the Shunt routes that allow comparison of currents in max demand operation.
You can monitor output voltage to see if there is a glitch (voltage droop) if the output demand increases. But this ONLY works if the CCS is working.
If the CCS is not working then a small voltage drop at the output increases the Vdrop across the CCS resistor and the CCS resistor is now effectively regulating for the output demand.
Last edited:
but what is CCS resistor Vdrop as the input voltage is changed?
eg.
with the PSU supplying 15.6Vdc and Vdrop =1.570V
add a resistor before the regulator. 2r2 will drop Vin by ~0.3V to ~15.3Vdc. what is the new Vdrop?
Add second 2r2. Vin= ~15Vdc? new Vdrop = ?
add a 3rd 2r2. New Vin = ~14.7Vdc new Vdrop = ?
Plot them on a graph.
If there is a knee in the Vdrop plot then that lets you define the minimum Vin for Proper CCS operation.
If you are already below that knee then you must change your Vin upwards to identify the Minimum Vin.
If you have a small Variac then you can vary Vin by the turn of the adjustment knob.
eg.
with the PSU supplying 15.6Vdc and Vdrop =1.570V
add a resistor before the regulator. 2r2 will drop Vin by ~0.3V to ~15.3Vdc. what is the new Vdrop?
Add second 2r2. Vin= ~15Vdc? new Vdrop = ?
add a 3rd 2r2. New Vin = ~14.7Vdc new Vdrop = ?
Plot them on a graph.
If there is a knee in the Vdrop plot then that lets you define the minimum Vin for Proper CCS operation.
If you are already below that knee then you must change your Vin upwards to identify the Minimum Vin.
If you have a small Variac then you can vary Vin by the turn of the adjustment knob.
Last edited:
NO.
If the CCS is not working properly because Vin is too low then adding a second CCS resistor makes the problem worse.
You should find out why the mosFETs are running so cool to be only very slightly warm. That alone tells me there is something amiss.
I have a standard DCB1, running at just 60mA. I cannot hold the irf240 with my fingers. They are too hot. Admittedly they each have 20C/W sinks.
You have more than double that current. V and/or A across the mosFETs must be exceptionally low.
By the way, with the ~150mAs the mosfets were absolutely cold to the touch after a few hours of listening.
I m ready to go higherI will leave it like that for a week or so to let it break in a bit and get a more solid idea of its sound while I work on the chassis and the volume control, and when everything else is more or less done I will hookup additional resistors.
I have some cheap 12R 5W ones which will be perfect for testing as you suggested
Some more first impressions after a few hours of low volume CD listening:
I have listened to a few CDs including some nice classic music by FIM (very nice mastering on most of them) and then some Classic Records. After that I switched to some plain rock and pop CDs. What got me by surprise was that some CDs that I considered really bad amazed me. Like the Metallica S&M. On the other hand some mediocre ones remained mediocre. Like the FIM Scotish Fantasy by Oistrach
The buffer seems really fast. The highs are excellent, the lows seem to go really low and are very very controlled and powerful.
I have some reservations about the mids and mid-lows but it could be because of the fresh mundorfs, the low volume and the cheapish pot.
About the soundstage, it is so freakishly wide even with that low volume. There are microdetails everywhere and the speakers have completely dissapeared. What was weird was that at some points the sounds were a bit fuzzy, almost as if the sound was out of phase. But this could also be because of the rough placement of the speakers.
My first day overall impression is that if I manage to get more focused soundstage and a bit more bite to the mids I will be crazy happy
In any case, the buffer is a keeper. And it had some really big shoes to fill since it replaced an Arc LS-25 and a Klimo Merlin
Check it up, tighten it up, boost it up, & there is more to hear.
To be honest I do not feel comfortable scoring the pcb to separate the psu from the reg. I will try to find a 15V transformer to see if I get a higher current. I have a 2x18 one but the electrolytics will be at their upper threshold with the filtered V.
I ve been listening to it for about 3 hours now and the mosfets are still at about finger temperature. The aluminum around the pcb feels slightly warmer that it does on its edges.
The buffer jfets feel a bit warmish too.
Are you sure about this? At 60mA shouldn't it be less than a Watt dissipation? And even with the 20C/W sinks shouldn t it be around 40-45 degs? Hardly a finger burning temp.
I ve been listening to it for about 3 hours now and the mosfets are still at about finger temperature. The aluminum around the pcb feels slightly warmer that it does on its edges.
The buffer jfets feel a bit warmish too.
Are you sure about this? At 60mA shouldn't it be less than a Watt dissipation? And even with the 20C/W sinks shouldn t it be around 40-45 degs? Hardly a finger burning temp.
If all is well you burn a mere 1.5W on each output MOSFET now. Just 0.84W on each CCS MOSFET. That metal base is big. If those red pads are Kerafol (ask Tea) then the case to metal thermal resistance is next to nothing. Those parts are 0.83C/W junction to case. Your test build is open also. They possibly glide merrily near room temp.
de-rate heatsink for Delta T.
Expect for a 1W into 20C/W at least a temp rise of 30Cdegrees, maybe even as high as 35C.
Ta = 20degC
Ts ~ 50 to 55degC
Tc ~ 51 to 56degC
Tj ~ 52 to 57degC
Ts & Tc @ >50degC is too hot to hold.
Note that for those temperatures, the devices should live for ever.
I design for the two worst case conditions
1.) Survivability if the output is shorted when mains voltage is maximum.
2.) Survivability if zero output current.
Expect for a 1W into 20C/W at least a temp rise of 30Cdegrees, maybe even as high as 35C.
Ta = 20degC
Ts ~ 50 to 55degC
Tc ~ 51 to 56degC
Tj ~ 52 to 57degC
Ts & Tc @ >50degC is too hot to hold.
Note that for those temperatures, the devices should live for ever.
I design for the two worst case conditions
1.) Survivability if the output is shorted when mains voltage is maximum.
2.) Survivability if zero output current.
Last edited:
The arduino controller is almost ready.
I used an encoder with button for the manual rotary control allowing 20 steps from 0 to 100%, and an apple remote for IR allowing 100 fast steps. The arduino then controls 2 100K DS1804 digital pots. A led is also dimmed to give a visual indication of the current volume level.
The only things remaining are to replace the second pot (used a 10K accidentally) and then properly implement the muting and test it more.
The digital pot will then control a lightspeed clone. The balance pot will be replaced by a trimmer to allow for more accurate centering since I always use it centered while listening anyway.
I plan on using a separate small toroid for the arduino, and power lightspeed with a single 7805 (no filter caps) from the Vout on the DCB1.
Any thoughts?
I used an encoder with button for the manual rotary control allowing 20 steps from 0 to 100%, and an apple remote for IR allowing 100 fast steps. The arduino then controls 2 100K DS1804 digital pots. A led is also dimmed to give a visual indication of the current volume level.
The only things remaining are to replace the second pot (used a 10K accidentally) and then properly implement the muting and test it more.
The digital pot will then control a lightspeed clone. The balance pot will be replaced by a trimmer to allow for more accurate centering since I always use it centered while listening anyway.
I plan on using a separate small toroid for the arduino, and power lightspeed with a single 7805 (no filter caps) from the Vout on the DCB1.
Any thoughts?
Hehe yeah at first I was thinking to build something like the one in the Buffalo thread in avclub.
But since I won' t need all the DAC stuff I decided that there is no point. The LCD will be an overkill. Just 3 leds for power, 2 sources and volume indication.
I know what you mean about the LDRs... I initially was planning to use separate pot quadruplets for each LDR (100K+1K) to perfectly control the LDR values per step, but it was really a chore... Not to mention the 16 pots (48 digital pins would be required....)
But since I won' t need all the DAC stuff I decided that there is no point. The LCD will be an overkill. Just 3 leds for power, 2 sources and volume indication.
I know what you mean about the LDRs... I initially was planning to use separate pot quadruplets for each LDR (100K+1K) to perfectly control the LDR values per step, but it was really a chore... Not to mention the 16 pots (48 digital pins would be required....)
I bought a DMM with temp measuring. The cold metal plate was 25,5C before turning it on. Same with the mosfets.
After ~1 hour the fets were at about 31C.
Now I have just boosted it a bit attaching 2 additional 12R @5W resistors on each rail bringing it to ~3,75R.
Here are the measurements
Vout: 9,95V and -9,99V
CSS: 1,412V and 1,391V
Offset: -1,3mV and -0,9mV
and about 20 minutes later
Vout: 9,9V and -9,96V
CSS: 1,411V and 1,392V
Offset: -1,3mV and -1mV
This translates to ~370mA
Still a bit lowish, but it seems to work nicely.
First impression was that the change was very noticeable. The music is more dynamic. The entire spectrum is much more solid and well defined. Highs are very clear, mids and mid highs have more bite and bass is louder and better controlled. And voices are starting to .... holly cr@p
I am getting more and more impressed by it every day
After ~1 hour the fets were at about 31C.
Now I have just boosted it a bit attaching 2 additional 12R @5W resistors on each rail bringing it to ~3,75R.
Here are the measurements
Vout: 9,95V and -9,99V
CSS: 1,412V and 1,391V
Offset: -1,3mV and -0,9mV
and about 20 minutes later
Vout: 9,9V and -9,96V
CSS: 1,411V and 1,392V
Offset: -1,3mV and -1mV
This translates to ~370mA
Still a bit lowish, but it seems to work nicely.
First impression was that the change was very noticeable. The music is more dynamic. The entire spectrum is much more solid and well defined. Highs are very clear, mids and mid highs have more bite and bass is louder and better controlled. And voices are starting to .... holly cr@p
I am getting more and more impressed by it every day