Testing the Rod Coleman Fixed Regulator
As part of my 300B SE project design, I looked into various fixed bias arrangements and regulators. Rod Coleman has developed another fantastic circuit after the success of his DHT filament regulator which is now the preferred filament supply kit within the DIY audio community. After many years of refining the DC filament regulator, Rod came up with a clever design for fixed bias using the same concept: a gyrator and a temperature-compensated CCS. Instead of feeding a current through the DHT filaments, in this case the current is used to generate a clean bias voltage across a “bias resistor”. The bias resistor is bypassed by a capacitor as the high impedance loop formed by the regulator and the bias resistor is sensitive to pick up HF noise which could be amplified by our system.
The kit is of the same quality you would expect from Rod’s boards and very easy to build. It takes less than an hour to build the boards:
I used the following voltage doubler to test the boards. A cheap 15VA toroids can be used to generate +100V of clean raw supply. You can use alternative components subject to what you have at hand. Although with just one 470uF filtering capacitor is fine (C3) I added an extra which was a spare in my stock. R1 and R2 not just help smoothing the supply, but also act as a fuse as they will limit the maximum current:
N.B. The fixed bias resistor between grid and VBIAS node is missing in the above diagram.
The regulator is rock solid. I checked with my oscilloscope and didn’t pick up any HF noise. Good for a prototype on the bench. I used a bulky 20uF/250V Russian PIO for decoupling the bias resistor which ended up being a 4K7/5W wire-wound as I didn’t have a 3K3 at hand. I set the regulator at 20mA so bias was about 95V. My raw supply was 135V with one board connected to the voltage doubler.
The bias current can be adjusted with the trimpot between 20 to 25mA which gives a nice voltage range for bias purposes. You need about +20 to +40V headroom for the board FETs to operate at their best. This regulator is suitable for low voltages to high bias voltages without any problems. Just a careful raw power supply design and build is needed.
The raw supply is really quiet. I added a second 470uF/160V at the end instead of one 470uF capacitor. After my initial tests I was expecting a higher rejection of the raw supply ripple (e.g>70dB) for this regulator. However, my workbench was too noise for me to measure this level of noise rejection. To give you an idea, we are talking about that my workbench noise level (with supply disconnected) is about 90uV (see next diagram), then the 120.4uV of inband noise is very low! Just 30uV from the overall noise in my workbench! Yes, if you shield all this thing around, you will probably get better results. For a high-impedance node is really good in my view! I had the board with alligator clips and point to point raw supply connected next to it, so no shielding here.
This is the inband noise level of my workshop when I measure the noise across the 4K7 bias resistor with the power supply turned off:
First tests were successful. I just need to build the 300B stage to try this and listen to the regulator. The proof is in the pudding as they say.
I was thinking of using one of the 4-wire Kelvin film capacitors to bypass the bias resistor as it will be very useful to improve the HF rejection given its low leakage inductance.
The regulator proved to be very effective to provide a stable and reliable fixed voltage whilst isolating the grid from the mains AC noise thanks to design of this regulator. A cheap low power toroid transformer is good enough to provide effective raw supply and its small size given low power requirements keep the parasitic coupling capacitances at low levels which are not impacting the circuit (note this is not the case in the DHT regulator due to power levels the EI transformer with split bobins is preferred).
Hope Rod releases this new board shortly as it’s a very promising design which I’m sure it will take fixed bias designs to the next level.
Hi Ale,
All looks very good. It would be great to compare the sound of a filament bias DHT ( I call this a quasi fixed bias because the circuit takes care of both bias and filament supply) to the sound of fixed bias. The question would be where to compromise. Is it worth to deal with a lot of dissipated power( heat ) in the case of filament bias if the sound is only slightly better? Is it worth to raised the complexity and cost with fixed bias by adding one separate supply? I believe that the sound resulted is more important than everything else…
Obviously, there are not a lot of DHT’s where filament bias can be done without needing a power plant next door..
As I said, proof is in the pudding and I’ve not listened to this regulator yet. This is a fixed bias regulator and has nothing to do with the filament bias arrangement.
I always preferred the sound of fixed-bias versus capacitor bypassed cathode implementation. Filament bias is great but not worth the challenge unless you are using 4P1L or similar. Believe me, I can fry an egg with my my 46 in filament bias 🙂
Ale,
Totaly agree that there are few valves where filament bias can resonably apply. I play a lot with 4p1l. In a PSE amp the filament resistor dissipates about 27W, or 54 W for 2 channels. It gets pretty hot…
Hi Ale,
Coleman regulator for fixed bias is very promising.
Can the standard coleman filament regulator handle the -135v supply?
Have you noticed any audible improvement in sound with this supply against your standard mosfet negative bias supply?
You have bypassed the 3k3 Bias resistor with a 20 uF PIO cap to ground.
Does it not bypass all the audio signal comming to the 300B grid to ground?
Do you install this supply between the 100k to 330k grid bias resister
and ground? Have you to change the 1k grid suppressor resistor to a different value if this supply is used?
Can this method be used with inter stage transformers and also grid chokes too?
I do not agree with Raductar in his previous comment where he calls this a
quasi fixed bias supply because you have also to use an additional filament regulator for the 300B with this grid bias.
Rod’s transmitter regulator could be used to filament bias the 300B if you
are prepared to dissipate about 170 watts of power and heat in the filament bias resistor.
Please publish your 300B circuit so that we could clear our minds about all these questions and agree with you more so.
We value your hard work very much and thank you for doing all the experiments we would have loved to do ourselves if we had the time, knoweledge and the resources.
Reding your blog gives us the same satisfaction we would otherwise have got doing these things.
The fixed bias point is connected to one of the IT secondary ends. If you are coupling it via capacitor then you will connect it to the grid resistor directly. The 3K3 acts as your grid resistor so you will connect this directly to the grid.
The 20uF capacitor is part of the signal path, so will have to listen to confirm impact in sound
A lot more tests to be done.
Again, this is a beta board from Rod, so we should expect some final refinements before he release this to the public… (i’m just a guinea pig)
Hi,
I have tried this idea in my GM70 amp – a cascode mosfet CCS creating a voltage drop over the grid resistor (RC coupled amp), and it sounded better than the classic fixed bias.
Rod’s module should be good too, but that big bypass cap – you surely can’t connect it directly to the grid like in your schematic, or it will short the signal to the ground.
Hi Vlad,
Thanks for your comments. Yes, it should be much better than traditional fixed-bias arrangement. The high impedance path with the raw supply is key and also Rod’s design has temperature compensation to improve bias voltage stability.
You are right about the schematic. I missed out the grid resistor. I was thinking on my circuit design which uses a transformer to couple signal from driver. In this case will be connected to one end of the secondary and the other end to the 300B grid. If capacitor coupled then the fixed bias resistor (circa 30-50k) should be placed. I will amend diagram when I get the chance.
The capacitor has to be around 10uF for proper filtering of HF. Bear in mind that the high impedance of the regulator makes it a great candidate to pick up HF noise. I haven’t tried 1-10uF range, Rod recommended 10uF. Should do the tests and see if any HF is audibly present. I feel much better with a smaller cap here though.
cheers
Ale
UPDATE ON FIXED BIAS BIAS REGULATOR (sent from Rod, 2016-08-11)
..yes, I found a way of designing them that’s even quieter, and more stable!
The design is making its way onto PCB now.
So I hope to have them available soon – Please remind me in a few weeks, I will have an update
Hi. Is the fixed bias regulator board available to buy yet? I can’t find it on the internet. Thanks.