CCS in power supplies

The use of CCS in HT power supplies is well known, however generally misunderstood why it can be a good addition to some circuits. There is an excellent article from Gary Pimm which has been lost and luckily I found it on my archive. Here it is in case you haven’t read it. Worth it as a refresher or for anyone who is new to the subject:

Gary not only introduces the CCS, mu-follower hybrid topology using FETs, but also covers a key aspect of the current flows in a driver (or preamp or output stage) with the signal and supplies. A lot of us have been in the quest of optimising the “sound fingerprint” of our power suppliers. Not just the HT, but also the filament supply which as we well know it affects big-time the DHT stages.

I want to focus on the aspect mentioned by Gary around the current control and separation. This resonated with me early on my hi-fi work and immediately adopted a few of these concepts and ideas on my designs. Let’s start with the typical stage. This one is DHT, simply for the pleasure of it. However, it can be any stage you want:

You can see in the diagram above a few set of currents flowing in our circuit. The blue traces are the input and output signals. The green is the filament supply and the red is our HT supply. We can clearly see why the HT supply contributes to the “sound” of our stage as it’s on the path from the interstage transformer (IT) to ground as return loop. As Gary points out, we want to control and prevent these currents from overlap and interact with each other. Ideally the “noise” of the PS we want that to be out of the signal loop. The filament regulator (e.g. Rod Coleman’s regs) provide that level of isolation on the filament side. However there are ways we can improve what’s going on on the anode:

In this second diagram we can see the introduction of the VR to add some level of regulation to the HT voltage. This works ok, however the purple plot show the return path which still exist and co-exist between the signal and the power supply. We can do better than this:

In this final diagram, we can see what the introduction of the CCS brings to the table. The supply and signal currents get isolated or moved apart to a larger extent. This is thanks to the CCS’ isolation (or PS noise rejection capability) which is a real benefit we exploit here. Yes, there is never a free lunch. The CCS needs at least 25V across it to operate effectively, so there is power dissipation to be considered and also the outstanding isolation of over 100dB taking place at LF is not that great at HF. However, the benefits outweigh massively the costs here. A win-win scenario. The CCS needs a low impedance node at the output to avoid stability issues. Luckily the VR with its comp cap (C1) provides that low impedance needed as well as a defined voltage. Also it looks amazing! Who doesn’t like the colour and looks of the VR valve glow!

The cascoded depletion MOSFET pair works extremely well for most of the cases, so no need to enter into more complex arrangements to improve the HF noise rejection. You won’t need it in most of the cases.

The above topology and its variations is a great arrangement to introduce to amplifier drivers or pre-amp stages. In particular when using choke or transformer-coupled arrangements as the PS noise rejection is lower compared to a hybrid mu-follower topology (i.e. using a Gyrator PCB). I implemented this arrangement with the 26, 01a and other valves with outstanding success.

Recently we had a conversation around this topic with my friends from Azzolina Audio. They have implemented this arrangement using my CCS and SiC bias PCBs in their outstanding Tau amplifier. Have a look at their website here.