AC filament power has a number of advantages, not in the least that it is a simple good sounding solution. But a view on the noise spectrum is a real eye opener indeed. Generating AC to solve it is cumbersome and you will still have the power transformer in the cathode circuit and still have intermodulation of the AC 50/60 Hz.
Why not rectify whatever you have and filter the crap out of it? Then go current regulated DC.
This is where I'm going. A seperate chassis with LCRC filtering using HF construction techniques (drop the power transformer through a metal plate into a HF-tight metal enclosure, use very short leads, HF snubbers and feed through capacitors in a segmented bulkhead like setup). Finish it off with a voltage regulator. Then feed this clean power to the audio chassis and power a current source that is directly mounted on top of the filament pins.
Why not rectify whatever you have and filter the crap out of it? Then go current regulated DC.
This is where I'm going. A seperate chassis with LCRC filtering using HF construction techniques (drop the power transformer through a metal plate into a HF-tight metal enclosure, use very short leads, HF snubbers and feed through capacitors in a segmented bulkhead like setup). Finish it off with a voltage regulator. Then feed this clean power to the audio chassis and power a current source that is directly mounted on top of the filament pins.
Hi guys,
I am just building an KT88 PP amp and I have already made my decision about filament heating when I read these posts as well as Gerits excellent site about filament supplies.
The strategy that I wanted to follow is:
- For driving tubes (5842) I will use DC current regulated filament supply. I will follow Gerits proposal to filter DC with the choke before regulating it with LM317.
- For KT88 output stage I will use AC filament supply. Here I am installing a separate transformer (230V/ 6,3 V, 5 A) by Hammond.
In addition though, I am putting on the mains primary side LC mains filter, hoping in that way to get better AC picture on the KT88 filaments.
Never done this before, though. Any comments on this strategy ? Any experiences already with this or similar approach ?
I am just building an KT88 PP amp and I have already made my decision about filament heating when I read these posts as well as Gerits excellent site about filament supplies.
The strategy that I wanted to follow is:
- For driving tubes (5842) I will use DC current regulated filament supply. I will follow Gerits proposal to filter DC with the choke before regulating it with LM317.
- For KT88 output stage I will use AC filament supply. Here I am installing a separate transformer (230V/ 6,3 V, 5 A) by Hammond.
In addition though, I am putting on the mains primary side LC mains filter, hoping in that way to get better AC picture on the KT88 filaments.
Never done this before, though. Any comments on this strategy ? Any experiences already with this or similar approach ?
I agree that this is the most practical solution. It's just that if you have a clean AC source all your gear will benefit from this. I started to think along this route after noticing the improvements my mains distribution block had on everything connected to it (mostly because of the very solid ground connection).
This would also be my choice for an ultimate heater supply setup, although the footprint is becoming rather large.
dejanm,
Thanks for the kind words.
Be sure to also take a look at John Atwood's DC filament supply test
For indirectly heated tubes you could also try a SMPS (properly grounded). I was quite surprised by the spectrum of a laptop SMPS followed by a VCCS, it was absolutely clean from 20Hz to 20KHz.
Gerrit,
Sorry, but what is: SMPS ? Any schematic of it ?
BtW - I've read that excellent article that you suggested on your web site. That is also the reason why I am going to put a choke for filtering filament supply for driving tube.
Sorry, but what is: SMPS ? Any schematic of it ?
BtW - I've read that excellent article that you suggested on your web site. That is also the reason why I am going to put a choke for filtering filament supply for driving tube.
SMPS = Switching mode power supply, the kind you find in computers and notebook adapters.
This is not my preferred choice, but it is a cost effective solution with a very small footprint.
Two suggestions for the traditional approach:
1 Use common-mode chokes.
2 If you can afford it, get Jensen 4pole capacitors (~€40). They really clean up the HF.
Thanks Gerrit for this information. I will have to try it - I have some decent and quite expencive Laptop supplies ...
Concerning Jensen - I've already bought Jensen 100uF, 500V for HV supply. So these best Jensens are probably not an option any more but this info is valuable for some future projects. The same goes for chokes - I will use them in the future.
You didn't comment the idea of having mains filter on the primary side and then to use AC for KT88 filaments .... I am having in mind the following mains filter implementation:
http://www.lampizator.eu/AC FILTER/SILK/SILK.HTML
Concerning Jensen - I've already bought Jensen 100uF, 500V for HV supply. So these best Jensens are probably not an option any more but this info is valuable for some future projects. The same goes for chokes - I will use them in the future.
You didn't comment the idea of having mains filter on the primary side and then to use AC for KT88 filaments .... I am having in mind the following mains filter implementation:
http://www.lampizator.eu/AC FILTER/SILK/SILK.HTML
I have no experience building them, I just buy RF filters. In my mains block I used RF filters only for the 'dirty' devices (TV, Behringer stuff etc.). There are no filters in series with the amps.
As far as I understand it, the filters will have little effect on the mains harmonics. Their main purpose is to filter out RF interference.
Has anyone used or examined floating plus/minus DC supplies for filaments? See circuit below. The series resistors far right represent an 813 filament (5 amps at 10 volts) with a fictious centre split. Just left of them are two 10 ohm resistors tying either end of the filament to ground for an effective 5 ohm cathode-to-ground impedance. Each resistor dissipates ~2 1/2 watts, increasing the filament circuit power dissipation 10% for an 813, worth the tradeoff for a low and linear cathode impedance.
The hash chokes are off-the-shelf uH Hammonds. The reasoning behind floating the DC supply is that since it's balanced across the filament with no low impedance path to ground the primary audio path to ground is via the 10 ohm resistors. The large DC smoothing caps are effectively no longer in the audio circuit. Spice analysis of the voltage to ground at the fictious filament centre, which should represent the common 'audio' voltage at either end of the filament to ground, is below -200 dB across the band. The result is of course a Spice artifact caused by perfectly matched ideal devices, the real induced noise will obviously be higher but since the filament supply floats relative to ground the supply noise should remain well balanced either end of the filaments and the performance remain high.
uH chokes are shown because I'm trying to make use of a 10 VAC Hammond transformer on hand and can't afford high DCR drops. There's no reason for example a 12 VAC transformer can't be used with higher value chokes for much better noise reduction with less peak current draw. Even so the spectrum from one filament end to ground is still excellent, falling into noise before 1kHz.
Look tenable?
The hash chokes are off-the-shelf uH Hammonds. The reasoning behind floating the DC supply is that since it's balanced across the filament with no low impedance path to ground the primary audio path to ground is via the 10 ohm resistors. The large DC smoothing caps are effectively no longer in the audio circuit. Spice analysis of the voltage to ground at the fictious filament centre, which should represent the common 'audio' voltage at either end of the filament to ground, is below -200 dB across the band. The result is of course a Spice artifact caused by perfectly matched ideal devices, the real induced noise will obviously be higher but since the filament supply floats relative to ground the supply noise should remain well balanced either end of the filaments and the performance remain high.
uH chokes are shown because I'm trying to make use of a 10 VAC Hammond transformer on hand and can't afford high DCR drops. There's no reason for example a 12 VAC transformer can't be used with higher value chokes for much better noise reduction with less peak current draw. Even so the spectrum from one filament end to ground is still excellent, falling into noise before 1kHz.
Look tenable?
Attachments
Greetings Everyone, Good food for thought here. There can be two problems with usings an SMPS for tube heating. I have two amps with SMPS heaters - not DHT though. One using 4 EL84s (series parrallel arrangement) is OK as the load is less than half the rating of the SMPS. (about 1.2 amps out of 3.0 rating). The other uses 6L6GCs and was close to 2.3 amps on the same type of SMPS. It had touble starting up. The start up load appeared as a "short" to the SMPS. One brand would start another wouldn't. Worse though, was that when running, the one that did start was putting out high frequency (about 60K HZ) switching spikes. These found their way into the signal path and could be seen at the outputs. I ended up replacing the SMPS on that amp with a conventional DC supply.
- Status
- Not open for further replies.