I do not agree. With few easy tweaks in the PSU, including fixed bias, it can work fine even without any loop feedback in UL operation with the right devices. The circuit might be improved with a CCS for the LTP tail however I found that this is not necessary. Recently I have made one and I got about 12W at the clipping (5% THD) from a single pair of PCL82's in UL operation without exceeding any max rating! After fine balancing, at 1W THD was about 0.3% at 50 Hz and less than 0.2% at 1KHz.
With quality NOS PCL82's and the right speakers it sounds amazingly good.....
Last edited:
The 5-10 already is UL.
CCS for the LTP tail is not really needed for ECC83, as it has high mu. Low mu LTP, as some people prefer, does need a CCS.
Fixed bias may help a bit, but not really necessary for playing normal music. Fixed bias will improve measurements by fixing the bias at the optimum point, but real music won't shift cathode bias by much so OK.
CCS for the LTP tail is not really needed for ECC83, as it has high mu. Low mu LTP, as some people prefer, does need a CCS.
Fixed bias may help a bit, but not really necessary for playing normal music. Fixed bias will improve measurements by fixing the bias at the optimum point, but real music won't shift cathode bias by much so OK.
Yes : Spendor SA1.
I haven't used the CCS. The front end is the original one with 1/2 ECC83 and the 2 PCL82's triodes in LTP.
The fixed bias has been made using the back bias method which doesn't need any additional supply. It's just a bit more...clever! You just move the cathode bias resistor in an another position and add some CRC filtering. You might need about 20 V more to preserve the supply for the front-end.
The other changes in the PSU are mainly related to capacitors that are bit small as usual in old designs. Nothing really different. So I have to disagree again, it is a Mullard circuit as this mainly refers to the front end. Tweaking a bit the PSU and the bias of the power stage is not what I would call a different design. It would be quite a ridicolous claim.....but I know this is common practice!
Last edited:
No. It is pentode.
http://www.r-type.org/articles/5-10-b1.jpg
When Original Mullard 5-10 is used without GNFB, its performance collapses.
Pentode connection was just one option:
Mullard 5-10. Ten Watt Amplifier
The original Mullard 5-10 could be both pentode and UL. No claims about new designs to be made....
I used 50% UL for G2. It works really well and the transformer design is the same as the one for pentode or triode with the exception of the taps brought out at 50% of the primary turns.
Last edited:
You are (almost) correct. I had not noticed that the original circuit was pentode.artosalo said:
However, Mullard offer 'distributed loading' (i.e. UL) as an alternative for 5-10 in their book and this is the way it is often built. I just assumed, wrongly, that it was always UL.
One other way to get cheaply a fixed bias for the PCL's 82 with no additional expenses for the PSU transformer could be using the filament secondaries if one doesn't need more than 20-21V bias. It can be derived in parallel putting a cap for DC decoupling and then a simple SS bridge rectifier and filtering to get about -22V max bias. I have to try this, yet.....
Also PCL'82 can be used at 300V anode and G2 voltage with zero problems, just like the ECL82. I have done this many many times. I think the 250V max rating was a consequence of series operation in televisions and this is not the case of a separate HiFi amp.
Also PCL'82 can be used at 300V anode and G2 voltage with zero problems, just like the ECL82. I have done this many many times. I think the 250V max rating was a consequence of series operation in televisions and this is not the case of a separate HiFi amp.
This is the frequency response of UL-connected 5-10.
When no GNFB is used, the response is hopeless.
Even with 26 dB GNFB, the high end falls too early to be "optimum".
With pentode connection all these are even worse.
When no GNFB is used, the response is hopeless.
Even with 26 dB GNFB, the high end falls too early to be "optimum".
With pentode connection all these are even worse.
An externally hosted image should be here but it was not working when we last tested it.
I think the 14W amp which that graph refers to is the one with pentode output. 14W is the power at the clipping.
The 20% UL as used in that circuit clips at 11W.
As I already said I got less than 0.2% at 1KHz at 1W with no feedback (50% UL with PCL82). You should try to see by yourself.....
I don't think that response is hopeless, it is -1dB down at 5Hz and 25Khz at 1W and 25Hz and 20KHz at the nominal full power (i.e. 10W). Maybe you should listen to some music rather than worring too much about what happens out of the normal (music) audio band.....there could be something else more important when harmonic distortion is sufficiently low.
It also depends a lot on the actual output transformer and the quality of magnetic materials nowadays is a lot superior.
Last edited:
To say that an open-loop frequency response is "hopeless" merely exposes ignorance about feedback loops. The open-loop response has to be "hopeless" in order to ensure closed-loop stability.
NFB cannot be simply dialled-in or out, as some folk seem to imagine. It has to be designed-in or out. An amp designed for serious levels of NFB is bound to have a poor frequency response without NFB. People who don't like this should steer well clear of NFB, as they clearly don't understand it.
NFB cannot be simply dialled-in or out, as some folk seem to imagine. It has to be designed-in or out. An amp designed for serious levels of NFB is bound to have a poor frequency response without NFB. People who don't like this should steer well clear of NFB, as they clearly don't understand it.
This discussion seems to slip from original a lot !
The graph I attached is from UL version since the GNFB is 26 dB. See the whole article: Mullard 5-10. Ten Watt Amplifier
I said earlier that the frequency response is hopeless without GNFB after you had said
According to Mullard document the open loop response is some -11 dB at 20 kHz, which is not very HIFI.
So: The performance of Original Mullard 5-10 UL and pentode is very poor if the GNFB is not used (as originally planned).
This case can be dramatically different if the original concept is essentially modified, but then it is not anymore the Mullard 5-10.
I said in the beginning that :
And 45 replid that:
And I replied:
DF96, can you clarify what of above exposes (my) ignorance about feedback loops ?
Last edited:
I think you should read more carefully. The graph you attached is referred to the 14W amplifier which is PENTODE connection. The UL is only 11W. There is nothing bad about it especially considering the poor quality of magnetic materials at that time. It is a very old article, do you understand this?
Maybe you should build an amplifier befere criticizing. It could be MUCH better than your usual simplistic 6SL7 concertina driving EL34's that for is one of the worst sounding valve amplifiers ever, regardless of frequency response....
There is no collapse. Maybe your transformers are not good enough! With 50% UL operation the plate resistance of the output pentode is about 3K. Even if you have a poor 7K OT with just 50H inductance you will get -1 dB at 20Hz. A reasonably good 7K transfomer will have 100-120H with some 2-3 mA unbalance and 160-180H woth no DC unbalance.....
The high frequency behavior depends a lot on the geometry and I never got a transformer that rolls off at 15-20K. Maybe if you buy a cheap Hammond you will get this regardless of power stage operation.
Optimum design means NOTHING. You must optimize every SINGLE amplifier for best performance regardless of datasheet and general information....this doesn't mean you have invented anything. It is quite ridicolous claim, IMHO.
Last edited:
You seem to be (partly) right. The response plot seems to be from pentode connected version.
My original claim was that pentode connected version is not optimum.
You said to this, that you do not agree, and then you gave a long list of modifications that should be done, including UL-modification. Therefore we do not disscuss about same amplifier. I refer to original Mullard 5-10 and you speak about something completely different.
Concerning UL-version: If other components except OPT- as it seems - will be kept the same, then the open loop frequency response and THD are still far from HIFI-class.
If you want that we should disscuss about yor version, that's OK.
Just show the circuit diagram and main test results (THD,frequency response etc.). But please, leave away the comments about how it or something else sounds. I only want to disscuss about facts, not about your subjective experiences.
With ANY transformer the open loop frequency response is poor if original component values are used, because these determine the frequency response, not the OPT.
But with a good UL-OPT having low leakage inductance, the response can be improved by modifying the circuit, but again, then it is not original 5-10 anymore, and that is what I have been talking about all the time.
- Status
- Not open for further replies.