This amp uses electrolytics in the output coupling/ feedback stage. Would I have a better option there for the stock 220uF lytics? I tried raising those to the next available at 470uF, and it was not a good result; It dragged everything down to a lower output and more heat. Thanks for this response, I'm glad to read that I'm not alone in detecting many subtle effects from these arrangements...not all in my head! Sometimes I almost think of the capacitors as having an analogous effect on tone to reverb in different shaped rooms and auditoriums. The topology of the caps themselves seem to have some color? or maybe I took a bit too much acid, back in the day...lol
So in the AC section (ie where you have the high voltage). If this is your first time dealing with HV I would purchase a pre-made line filter and simply AC plug into that. There are IEC cords with a line filter built in (not the audiophile ones just things like schaffner). I suspect that the space you have available may not allow you to retrofit a filter IEC socket. The filter has a common mode choke setup and optionally other components to filter out noise coming in from the power. It won't remove it completely but it will make a good effort to attenuate it.
As to output caps - it depends on the headphones and what's driving it.
It's key to understand that the output coupling caps are there to (a) block DC and (b) provide 'isolation' between you and the tubes anode voltage (as per post 2).
For this reason the caps MUST be able to take the voltage of the DC + the AC peak on top (ie best take the B+ rail (max DC+AC) and add some additional such as 20% to cover any mains or transformer regulation. I couldn't see the voltage shown on the diagram but it will be at least +175V*1.414 = 247Vdc so best put at least 400-450V caps in there (you'll need to be aware of the possible maximum of your line voltage and the unknown mains transformer regulation they use). Other OTLs use the output from the cathode hence a lower voltage but this design uses the anode hence high voltage - and I can tell your diagram is incorrect here because theres no anode HV connection to the power supply.
Caps such as Nichicon MUSE ES are NOT suitable for this (in C8/C9) as they only go to 50V. So don't copy other OTL cap suggestions blindly.
As to how much capacitance - that 6N6P has a 30mA max. If it's running at 15mA bias then it's got 10-15mA swing max. It's not a bank of four ec99s with 60mA each... hence trying to add too much capacitance may be detrimental to the sound. Essentially there's a relationship of current/change in time = charge, and from that you can then work out the rough uF. q=CV give you the capacitance for the voltage based on that charge. Now it would be integrated over the sine wave but you can simply make a static estimate - either square or do RMS over the 1/2 wavelength period. That's the complicated answer I think.
So assuming 15mA swing, 20Hz, and a 1Vpk swing..
0.015 amps * 0.05 seconds for 20Hz = 0.00075 columbs, so 0.00075 / 1 volt AC = 0.00075 Farads or 750uF max roughly..
So you could probably add some more capacitance but 400-450V caps get big fast.
As to output caps - it depends on the headphones and what's driving it.
It's key to understand that the output coupling caps are there to (a) block DC and (b) provide 'isolation' between you and the tubes anode voltage (as per post 2).
For this reason the caps MUST be able to take the voltage of the DC + the AC peak on top (ie best take the B+ rail (max DC+AC) and add some additional such as 20% to cover any mains or transformer regulation. I couldn't see the voltage shown on the diagram but it will be at least +175V*1.414 = 247Vdc so best put at least 400-450V caps in there (you'll need to be aware of the possible maximum of your line voltage and the unknown mains transformer regulation they use). Other OTLs use the output from the cathode hence a lower voltage but this design uses the anode hence high voltage - and I can tell your diagram is incorrect here because theres no anode HV connection to the power supply.
Caps such as Nichicon MUSE ES are NOT suitable for this (in C8/C9) as they only go to 50V. So don't copy other OTL cap suggestions blindly.
As to how much capacitance - that 6N6P has a 30mA max. If it's running at 15mA bias then it's got 10-15mA swing max. It's not a bank of four ec99s with 60mA each... hence trying to add too much capacitance may be detrimental to the sound. Essentially there's a relationship of current/change in time = charge, and from that you can then work out the rough uF. q=CV give you the capacitance for the voltage based on that charge. Now it would be integrated over the sine wave but you can simply make a static estimate - either square or do RMS over the 1/2 wavelength period. That's the complicated answer I think.
So assuming 15mA swing, 20Hz, and a 1Vpk swing..
0.015 amps * 0.05 seconds for 20Hz = 0.00075 columbs, so 0.00075 / 1 volt AC = 0.00075 Farads or 750uF max roughly..
So you could probably add some more capacitance but 400-450V caps get big fast.
Thanks, that is a great answer! I am fine with HV...worked with it in many forms from Tesla coils to automated machinery using 700V. DC and 3ph 480. I appreciate your concern, none the less! Also thank you for the math.
I have spotted a mistake - that 1/2 wave means that's 750uF/2 or 375uF without RC or frequency derating effects.
The problem getting upgrade-itis is that the chassis suddenly shrinks 😀
As last point - you could always have a larger HV electrolytic then bypass with a foil (I assume the situation with high AC to low side doesn't cause a reverse polarity). PP foil WIMA do 630Vdc/400Vac that would work for not much investment. The issue is size. I like the FKP1 once it's had a few hours on the clock but there's also other variants that aren't going to cost the earth compared to the boutique HV caps. However down the path of cap rolling there be dragons 😀
Last edited:
Impedance really, if the larger cap electrolytic isn't supporting the higher frequency then adding a foil cap in parallel helps address that.
own personal opinion in that the output cap with a bypass is faster but yes the bypass alters the filter behaviour.
on a different note... one of my earliest memories with my dad, who was an EE, was making huge caps out of window glass and tin, for a big tesla coil we were building. It was the best lesson ever in how a capacitor works because you could even see it working. Each plate was about 18" square, in a vertical rack.
you could use a filter cord rather than replacing the socket as I think you will have space issues. Having a iec cord with a filter built into it means it’s out of the case constraint.
Hi Nick, what do you think the lowest voltage rating I could use on the 220uF output coupling caps? It is part of a neg feedback loop, too. the cans are rated at 200V but I want to try another brand, and highest is 50V. I think it should be fine but what's your thoughts?
you can stack the caps in series and use resistors to balance the charging to give higher voltage.
you must support the b+ rail voltage to ground in case the plug us shorted. It’s the only thing isolating you.
you must support the b+ rail voltage to ground in case the plug us shorted. It’s the only thing isolating you.
I am looking for Elna Silmic II 250v caps...any idea where to look? Besides Mouser and Digikey, who do you go to for quality audio caps? thanks.
I stick to known component sellers - mouser, digikey, farnell, rsonline. Too many forgeries.
I found a great, simple improvement to the Little Dot amp: lose the input bypass caps! Totally unnecessary and bad for the sound. I removed them and just shorted their terminals and the sound is much smoother and clear. No need for these that I can find. My SVT does not have them, and so many others do not, so I found no good reason to have them here. Best cap is no cap!
On a different subject, I wish I had the ability to flip the phase of the input signal. Can someone help me understand where the wisest point in the circuit would be to do this, and how? The common ground is confounding me.
Thanks
On a different subject, I wish I had the ability to flip the phase of the input signal. Can someone help me understand where the wisest point in the circuit would be to do this, and how? The common ground is confounding me.
Thanks
Not possible, this is inherently an inverting circuit. Use an input transformer.
Otherwise, reverse the polarity on both of the headphone transducers instead.