And to put it in context, the offboard power supply goes EZ80 - 560R dropper resistor - 47uF -20H - 100uF (line stage supply for another RC stage taken off here) - 25H - 100uF to the phono stage power input. Should be sufficient overkill in any case to kill any ripple even with a cascode stage with zero PSRR.
100hz would seem like summat wrong with the supply. Cap swapping and measuring things.
Successive stages of filtering can indeed eliminate ripple which has propagated ‘along the wires’ (for want of a more accurate phrase). But if, in the first stage of the circuit, you have very short, high-current cap recharging pulses then they can generate noise which propagates radiatively i.e. ‘through the air’ or by inducing stray currents in the chassis and this can get into the signal path by bypassing the subsequent HT filtering stages. The only solutions are i) not to generate the noise in the first place, by keeping the peak currents under control and ii) to confine it to the space where it’s generated.
The PSU is a good metre away from the phono stage as it is sited now and there is a 560R in line with the output of the EZ80 (to drop the overall voltage to what is required and also as a plan to limit the current through the first cap). That bit hasn’t changed from when the hum wasn’t there.
PSU is bottom left, phono stage is top right.
EZ80s can be a bit prone to heater-cathode leakage (they were introduced with 6.3V heaters, when 5V was much more common, to allow them to use the same heater winding as all the other valves in 5-valve radio sets, but this can put a lot of strain on the h-k insulation). This can couple pulsed 100Hz noise into the heater circuitry of all the other valves. I’ve seen it ‘in the flesh’ in a Trilogy pre.
Does the EZ80 have its own heater winding ? If not, can you rig up a temporary isolated 6.3V supply for it just to eliminate that as a possible route for the noise ? I realise this is in direct contradiction of my last statement about the circuit change being the most likely cause of the new hum, but annoying coincidences do happen sometimes too.
It does. It uses the 6.3V winding from the Hammond trafo that also does the HT, there are two other regulated heater supplies (one referenced to signal ground, one elevated to c. 50V for the cascode and the cathode followers) from a separate transformer.
Surely the only substantive change you made is going from electrolytics to films? - so revert that change and see if the problem still exists.
Done it.
Still hum.
Dry joint or similar dodgy connection?
Put it in the bin.
Moved some wires around and hum substantially reduced.
This needs addressing a bit more though.
Later today I will mostly be doing smoothing. There is a naughty saw tooth on the psu rails. As a first lash up I am happy so far.
Musicy stuff comes out and nothing went smoky or Bangy.
6 Likes
Soooooo many possibles - any flyleads you can move about (rubber gloves/usual precautions, natch) and listen to see if its routing of any of them that’s affecting pickup?; my DAC is super-susceptible, partly due to being crammed into a teeny-tiny box.
Talking of DACs, got a brief loan of @rmsshipbroker ANUK DAC - something in there (loose laminations in a tx? dodgy choob?) is resonantly joining-in with the music! Not had the lid off yet (it’s not mine after-all), but intriguing.
Drill a small hole in the back. Fill the whole thing with potting compound. Usually shuts stuff up.
1 Like
Done 
Hope Steve’s pleased 
Slightly odd output stage - most examples have 2 x ECC82 - one per channel; didn’t dismantle but it appears to use 1 x ECC82 and 1 x 7044 (which I think might be ~=E182CC), with each triode of each valve driving a separate channel in tandem… Not 100% sure since the layout of the OP stage was different to any I’ve encountered and I didn’t want to dismantle everything to be sure. The alternative is someone using a 7044 in place of an ECC82, which is not a generally-accepted substitution…
No matter, it’s a really great sounding DAC, I want one…
2 Likes
There are, but one of the major precautions would need to be having two overly inquisitive KettleJnrs nowhere near them.
I may need to try and shoehorn one of Pete’s ground lift boards and a switch to see if that makes any difference (given there is some impedance between the preamp ground and phono stage via the 4 terminal reservoir cap that is now in the line stage), but that will also mean attacking the rear panel with a drill which is also not a small person friendly activity.
You need a shed.
As for ground lift, x-class cap + 10R to try the principle; seems good practice even if it doesn’t cure your hum.
The pin connections aren’t the same - the 7044 (and the E182CC) are different from the ECC82 from pin 6 onwards.
The 7044 has much the same gain as the ECC82 but can handle roughly twice the current and has roughly twice the transconductance. So I suspect each valve contributes one triode to each channel with, perhaps, the 7044 driving the cables as a cathode follower (high gm gives nice low output impedance).
1 Like
Cheers. Good to hear that doing things that way makes sense 
A lot of sense actually.
A classic problem is when designers use one of the various sorts of double triode ‘totem-pole’ circuit (mu-follower, SRPP etc). The cathode of one of the triodes will be at roughly the same voltage as the anode of the other one. Obvious problem: each valve’s two internal heaters are hard-wired together, so both will have to be at the same DC voltage. But the cathodes might be more than 100VDC apart. So there will be a good deal of heater-cathode voltage in one triode or the other or, sometimes, both. Vhk is generally A Bad Thing. Leakage. Even breakdown. Tears before bedtime.
But in a stereo unit there will be two double triodes. If you put one in each channel then both will suffer with Vhk. If you split each one between the channels though then one can have both its cathodes sitting close to ground and the other can have them both up at 100V+. Separate heater supplies cost a quid or two more and can be floated to whatever is the appropriate DC voltage. Sure, the upper cathodes will still flap about at different audio voltages (that’s stereo for you) but they will still be better off than if there’s also a DC offset to deal with.
OK, cross-talk. But it’s negligible in the valve compared with what it’ll be on your vinyl disc or in your listening room.
1 Like
Thanks, that’s really interesting. This looks an early unit, and the output stage is really nicely put together - all 1W tantalum R and Cerafine C - and all hard-wired together using the component leads themselves in an uncharacteristically-tidy minimalist way (for ANUK’s badgers). Later DACs all have one ECC82 per side - which given the issues you raise makes me wonder why they changed?