Filament DC supply for DHTs
Having played around for some time with DHTs with preamps (26,46,10Y,4P1L,71a,Cx371a and many others) and single ended amplifiers using DHTs as well, I thought it was worth mentioning a very critical point in a DHT stage: filament supply.
I’m not going to enter here in the academic debate about whether DC or AC is the way to go. Some orthodox audiophiles would say: AC filaments do sound better! They sound good to me but you know what? I never managed to get the hum out and in particularly whilst listening with my headphones it was a slight nuance which I sought to get rid of.
Despite AC filaments are very simple from a design perspective (i.e. transformer, wires and a hum pot), layout becomes a real challenge and sort of a black art. It reminds me of learning antenna and radio signals at University 🙂 Also sound is highly reliant on the transformer type and its isolation capabilities from the mains noise.
Many of the DHTs we like using in audio were designed for DC. But battery DC is not a practical solution, some have experimented and achieve really good results with them (have a look at VT52.com) despite the hassle of having the battery packs, charging circuit, etc.
DC regulators like LM317 were not designed for audio use. So don’t use them if you are looking to get good performance. The voltage dropout and AC impedance are markedly bad and it is noisy.
Rod Coleman came up with a great idea to isolate the filaments whilst minimising the ripple in a DC supply: a gyrator on the positive terminal and a CCS on the negative one.
Rod’s regulator design was done with the following in mind:
- Control current to a limit, to prevent damage to the filament caused by switch-ON surge (might be 8 or 10 x normal) over the years. Filament breakage is a big problem in modern 300Bs.
- Reduce ripple to the smallest possible level, since filament ripple modulates the anode current (intermodulation).
- Adjust filament voltage automatically, by gently adjusting a current source. The voltage control loop must have low bandwidth to prevent the dynamic impedance across the filament getting too low.
- Present a high impedance (especially low capacitance) to the power supply, to keep mains noise and rectifier recovery pulses OUT.
- Present a high impedance between the two ends of the filament. (Rod wasn’t entirely sure why this should be, and have never found any reasonable explanation. But if you doubt whether it is true, just try current-driven heating. Then connect 1uF across the filament – all the wonderful sound suddenly disappears!)
- Present a high impedance from the filament toward the power supply. This is to keep the anode current from capacitively coupling to Earth/Ground via the wiring and filament transformer.
So why using a DHT regulator then?
- Establish a high dynamic impedance between the Filament terminals. I don’t know if there’s some kind of micro-phasing effect, or something else, but it is certainly important. If you connect this circuit to a DHT and enjoy the wonderful sound for a while, then try putting a 1uF across the filament – the sound is degraded below AC standard.
- Minimise hum without interfering with the anode current. The gyrator is an open-loop regulator, while the CCS has feedback network which is not in contact with the filament. There’s a big difference. I tried circuits with opamps and other feedback methods, even rolled off as a 2Hz servo, but if the feedback network is in contact with the filament, you lose out on the sound. One can only assume that the anode current is being sensed by the feedback, partly, and causing interaction with the CCS pass transistor.
- Because the filament is the cathode, and the cathode is part of the DHT’s input [the grid-cathode voltage drives the DHT], imagine having the input circuit snaking around on long wires, picking up fields, and capacitively-coupling the input circuit to ground through horrible dielectrics (like PVC wiring). But that’s just what happens with AC-heat, and dc-rectified heating. That is why ac-heat sounds different with different transformers – although it always sounds bad compared to CCS heat.
- The CCS/Gyrator transistors buffer the filament from the filament trafo & rectifiers, and gives more consistent, and hugely better sound.
- Current limited control prevents big turn-ON current. My 300Bs measure 0.7-ohm cold, so they would see ~8-Ampere peaks at startup. That maybe why mine have already lasted nearly 12 years constant use, whereas filament breakage seems to be the biggest failure mode for 300Bs”.
How easy are they to use?
Not difficult, but won’t say dead easy. Stuffing all kit components in the PCB is very simple. You need to have some basic building experience with power transistors to ensure you mount them properly with the isolation bits, etc. The power transistors need proper sink and low thermal resistance path to ensure they dissipate the heat properly.
How do they sound?
Well, this is a personal experience. But if you check out there, everyone will report superb results. I personally tried Rod’s boards in many different DHT preamps: 30,30sp, CX112a, 01a, 4P1L, 71a, 46, 10Y and famous 26 of course. I’m also using them in my 45 SE amplifier. I can say that the results are astonishing. Sound is very clean, no hum, presence and tone are remarkable. Listening with my Grado headphones now is a pleasant experience. A happy man who also discovered this great contribution from Rod to the DIY audio community.
How do I get them?
I have no commercial relationship with Rod. I asked him whether I could share this info and Rod happily provided a contact email. These regulators are cheap for the value you get. The PCB board is of great quality and you will find Rod always available to help out.
Also there is a lot of information in DIYaudio about this. Have a look at this thread
Hope you find this info useful.