Hello!
I have someone's Radford STA35 in my possession for some simple repairs. Looking at the circuit for this amplifier, I can't help but think some of the resistor values, particularly around the phase splitter, could do with changing.
Here is the circuit
Since this amp is PCB based and the customer rightfully doesn't want excessive modification, I can't really do much to the circuit aside from changing values.
The anode loads for the phase splitter, R13 and R14 look WAY too low, and so does the tail. Ideal the tail should be "infinitely" long but 2k2 is just way down low! Aside from that, a gain imbalance is pretty much guaranteed thanks to the matched anode loads and short tail. Not to mention the distortion due to the steep 15k load.
To improve this stage I was thinking about putting a 39k in the position of R13 and 33k in R14, and changing R15 (the tail) to 8k2. These values are the same as the classic Radford STA15/25 circuit and at least seem to be more sensible. Simulating in SPICE reveals the overall gain of this arrangement ends up about the same, but with lower distortion and better matching.
Another thing bothering me was the extremely low cathode-anode voltages required for DC coupling on the input triode pair, around 45V for the top one and 35 for the bottom! The ECC81/12AT7 does not look pretty down this low, and I have an inkling that the ECC83/12AX7 might actually work better as a drop in replacement - initial SPICE sims seem to show lower distortion (Although another tube model results in the distortion being the same. This is why I don't just blindly trust the simulation results!)
Am I on the right track with this? I think the amp could benefit from these minor tweaks, especially to cut down any higher order HD that might result from the steep loads
Thanks!
I have someone's Radford STA35 in my possession for some simple repairs. Looking at the circuit for this amplifier, I can't help but think some of the resistor values, particularly around the phase splitter, could do with changing.
Here is the circuit
Since this amp is PCB based and the customer rightfully doesn't want excessive modification, I can't really do much to the circuit aside from changing values.
The anode loads for the phase splitter, R13 and R14 look WAY too low, and so does the tail. Ideal the tail should be "infinitely" long but 2k2 is just way down low! Aside from that, a gain imbalance is pretty much guaranteed thanks to the matched anode loads and short tail. Not to mention the distortion due to the steep 15k load.
To improve this stage I was thinking about putting a 39k in the position of R13 and 33k in R14, and changing R15 (the tail) to 8k2. These values are the same as the classic Radford STA15/25 circuit and at least seem to be more sensible. Simulating in SPICE reveals the overall gain of this arrangement ends up about the same, but with lower distortion and better matching.
Another thing bothering me was the extremely low cathode-anode voltages required for DC coupling on the input triode pair, around 45V for the top one and 35 for the bottom! The ECC81/12AT7 does not look pretty down this low, and I have an inkling that the ECC83/12AX7 might actually work better as a drop in replacement - initial SPICE sims seem to show lower distortion (Although another tube model results in the distortion being the same. This is why I don't just blindly trust the simulation results!)
Am I on the right track with this? I think the amp could benefit from these minor tweaks, especially to cut down any higher order HD that might result from the steep loads
Thanks!
Last edited:
I'd leave well enough alone, all sorts of issues can arise out of ill-considered and rushed changes arising out of a desire to improve things. The whole phase-splitter stage design concept IMHO is suspect, but you'll not fix it by just changing a few resistor values, and you could make it much worse.
You'll affect the loop gain, bandwidth, and operating point of the splitter stage which will require careful analysis to understand fully so that you can correct the compensation for stability into any load that this amplifier will encounter, and verify that the splitter is linear at all required drive levels, etc..
http://www.mif.pg.gda.pl/homepages/frank/sheets/127/6/6U8.pdf
http://www.mif.pg.gda.pl/homepages/frank/sheets/082/e/ECF82.pdf
Note that the triode's rp is around 6K, and it is operated at a pretty high current so the 15K load is not unacceptably low.
The splitter balance is dependent on a number of factors and just changing resistors could result in poorer overall balance and worse HF balance, changing the operating point of the pentode will also affect the gain of the triode based section due to a change in internal cathode impedance as a result of an operating point change. This was carefully and not accidentally designed by someone who knew what he was doing. I am not sure why this design was chosen, but certainly eliminates the effect of cmiller loading on the previous srpp driver stage. (Strikes me the design intent here was for as much bandwidth as reasonably possible.)
You might want to simulate the splitter design or better still measure its performance before drawing any conclusion about its linearity. Note also that it is direct coupled to the SRPP (low voltage at that) and there is probably around 20mA of current in the (short) splitter tail resistor.
You'll affect the loop gain, bandwidth, and operating point of the splitter stage which will require careful analysis to understand fully so that you can correct the compensation for stability into any load that this amplifier will encounter, and verify that the splitter is linear at all required drive levels, etc..
http://www.mif.pg.gda.pl/homepages/frank/sheets/127/6/6U8.pdf
http://www.mif.pg.gda.pl/homepages/frank/sheets/082/e/ECF82.pdf
Note that the triode's rp is around 6K, and it is operated at a pretty high current so the 15K load is not unacceptably low.
The splitter balance is dependent on a number of factors and just changing resistors could result in poorer overall balance and worse HF balance, changing the operating point of the pentode will also affect the gain of the triode based section due to a change in internal cathode impedance as a result of an operating point change. This was carefully and not accidentally designed by someone who knew what he was doing. I am not sure why this design was chosen, but certainly eliminates the effect of cmiller loading on the previous srpp driver stage. (Strikes me the design intent here was for as much bandwidth as reasonably possible.)
You might want to simulate the splitter design or better still measure its performance before drawing any conclusion about its linearity. Note also that it is direct coupled to the SRPP (low voltage at that) and there is probably around 20mA of current in the (short) splitter tail resistor.
Last edited:
Hi there, thanks for the reply
Keep in mind I didn't randomly come up with these alternative values - they are from a proven design with pretty much the same HT and grid input voltages to the phase splitter.
As I said in the OP, I have simulated both arrangements are the higher values did simulate with better performance.
Another issues is that with the 15k anode loads, the valve and resistors get hot, leaving significant scorch marks on the PCB, even with 3W resistors. The original Radford design works just fine with 0.5W resistors.
Keep in mind I didn't randomly come up with these alternative values - they are from a proven design with pretty much the same HT and grid input voltages to the phase splitter.
As I said in the OP, I have simulated both arrangements are the higher values did simulate with better performance.
Another issues is that with the 15k anode loads, the valve and resistors get hot, leaving significant scorch marks on the PCB, even with 3W resistors. The original Radford design works just fine with 0.5W resistors.
I'd say then that it is probably worth the experiment, scorched pcbs are never cool IMHO, and if the design you propose avoids this issue, and also improves linearity then there's another good reason. Leaves me wondering why they changed it in the first place. On reflection your proposals all sound worthwhile..
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.