Hopefully this post will be strictly “on topic”!
I will outline a project I undertook a few years ago to build a modern power supply for an R1155 receiver. Strictly the project was for a complete T1154/R1155 system, but I will describe just the receiver part of this. The psu was built – but I no longer have it, as it was done for a client and delivered. I may have the odd photo which I can share.
I started by establishing the requirement. Sure we have a rough idea of the basic volts and amps required – but I refer to more subtle things like:
- How much ripple is allowed on the HT rail?
- How stable should I make the HT voltage?
- Should the LT be DC or AC?
And I was aware that the original mains PSUs for T1154/R1155 were enormous brutes, with multiple chokes, etc. It seemed back in 1940 that the “requirement” wasn’t straightforward. And incidentally, I didn’t just want to get a R1155 “making a bit of noise” – I wanted it to work to the original spec. Trouble is that while the government would have definitely delivered a clear requirement spec to the original designers (Marconi?) that document is most likely gone for ever. I've never seen anything – though I suppose a search through the records at Kew might turn up something.
So I decided to reverse engineer the requirement…..by looking at what the original designers produced, and asking “How did that perform?” I didn’t have a working original T1154/R1155 mains psu (actually the pair) to hand, so I turned to the technical description of them – document enclosed (AP1186E Vol.I, Sec.6, Ch.4).
The basic amps and volts requirement is given as:
- HT: 210V DC at 110mA max (that's all for the R1155)
- LT: 7V DC at 13A max
The LT is misleading because most of that 13A is actually going to the T1154. The keying relay alone takes some 4A! The actual LT requirement for the R1155 has to be established separately by considering the valve heater current demands, and when that is done by recourse to AP2548A, it tots up to 4.67A with all the valves present.
So a nominal 5 amp requirement.
What about the 7V DC bit? Well the 7V is to allow for voltage drop in potentially long cables and various connectors, so that the valves actually get their nominal 6.3V, when the whole system is taking upwards of 10 amps or more. We can relax the 7V to 6.3V provided the cabling is short - i.e. a couple of feet or so of standard cabling. There is no actual need for DC LT in the R1155 alone. The DC requirement comes from the T1154’s keying relay.
So with the LT established as around 6.3V AC at 5A, lets turn to the HT. The original circuits are shown in the extract in my simulation results document, with values taken from the text of AP1186E. I then supplemented that info with some measurements on an actual original PSU, which was not working due to having duff metal rectifiers.
- Transformer gives 270V rms off load, with very roughly 50 ohm source resistance
- Chokes measure 11H (zero current) with series resistance of 105 Ω
- Other components near to nominal values. I had to guesstimate the capacitors’ ESR at 0.5 Ω
The point of all this detail is that the next step is to simulate the original power unit to see how the original design performs for stability and ripple.
Rather than trying to grapple with a full simulator (like pSpice) I use a
dedicated psu simulator, from “Duncan’s amp pages” – which is available for free download.
As it turned out, this simulator is severely taxed by aspects of this ancient psu – notably the metal rectifier. This may well be an instance where resorting to algebra and spreadsheets would give a better result – but I didn’t bother this time, mainly because I had no definite data on the metal rectifier anyway.
I have attached the original simulation schematics and results I recorded when doing the design work some five years ago. I’m aware that the simulation is a compromise – mainly due to the original metal rectifier, for which there is nothing in the simulator library and there is no data available that I know of. So I used a 4 x 1N4007 bridge then tried to simulate the increased series resistance of a metal rectifier by putting a series 100 ohm resistor before the filter stages. That value of 100 ohms was derived from reading early texts on metal rectifiers and their performance (Bell System Journal for 1953, “Selenium Rectifiers – Factors in their Application”, Gramels). Better ideas welcome!
The simulations used a number of conditions:
- 1. With a 8.8k load. This is the supply at startup, with the R1155 connected but the valve heaters cold. 8.8k represents the 50k resistor in the psu in parallel with the fixed resistive loads in the R1155 which are permanently in-circuit regardless of mode.
- 2. With a load of 2.8k giving an output current of about 75mA. This represents the receiver running in normal receive mode without any DF systems being used.
- 3. With a load of 1.9k giving an output current of about 110mA (design spec).
Results - Condition 1:
Load value = 8.8kΩ
Load current = 26mA
HT voltage = 232V
Ripple (Vp-p) =1.6
Results - Condition 2:
Load value = 2.8kΩ
Load current = 77mA
HT voltage = 214V
Ripple (Vp-p) =1.6
Results - Condition 3:
Load value = 1.85kΩ
Load current = 110mA
HT voltage = 203V
Ripple (Vp-p) =1.6
So the conclusion of all that is that HT rail is 208V ± 5V, and the ripple is pretty steady at 1.6V p-p. That remarkably stable HT rail is mainly due to the use of a L-C filter, with the addition of a background load (condition 1).
In my next post I will describe the modern version of this original circuit, and how well it performed compared to the original.
Richard