If currentbalance is OK you should either remove or short the cap, off course another thing if you later wish to balance currents electronically with a current mirror.
The cap serves no purpose as I see it, as this is a balanced design and NO LTP. Shorted no local feedback, open local feedback.
The cap serves no purpose as I see it, as this is a balanced design and NO LTP. Shorted no local feedback, open local feedback.
Last edited:
I initially used 20uF in a similar circuit with low mu tubes and had a bit of a problem with record warps so you might start to see bass extension with your circuit at not much over 100uf assuming mu of 16 for 5687 and 100H(?) for the OPT primary.
My circuit had tails about 50 Mohms long
My circuit had tails about 50 Mohms long
Last edited:
Marginal bass-boost with 10u but it is there. It peaks at 70Hz by 0,2dB with some guesstimates of the inductance values.
Shorting the cap gives the best low end extension and is -1dB at ca 15Hz. The other to are -3dB at the same frequency. But as I say the inductances are guesswork with the primary at 100H.
So removing the cap is no option. Either substitute with a big enough or short it.
I anyway get indications of a very good design(apart from the cap).
Shorting the cap gives the best low end extension and is -1dB at ca 15Hz. The other to are -3dB at the same frequency. But as I say the inductances are guesswork with the primary at 100H.
So removing the cap is no option. Either substitute with a big enough or short it.
I anyway get indications of a very good design(apart from the cap).
Short it=infinite cap. But from 470u it will not upset frequency-response.
Once again this is not a LTP. If you go for CCS or current mirror add a cap to each cathode to ground and no cap should between cathodes necessary. About the latter it might be a sonic benefit using your previous back to back arrangement, you decide.
Once again this is not a LTP. If you go for CCS or current mirror add a cap to each cathode to ground and no cap should between cathodes necessary. About the latter it might be a sonic benefit using your previous back to back arrangement, you decide.
Last edited:
.
Correct - when the current mirrors go in - with the back to back caps - it will be a LTP. This has been demonstrated by me on a number of designs with CCS. It is not necessary to implement a LTP in this design - but it is my decision to do so.
Accurate current balance is essential in this design and this is the only way to achieve it other than a DC servo (not going there). Shorting the cathodes will not maintain accurate current balance. I also suspect that the 10uf cap between the cathodes is going to go low enough because the output impedance is defined predominantly by the anode load rather than the cathode load. Until I build the current mirror version it will be difficult to say if there is a sonic penalty for the higher effective output impedance.
Shoog
Last edited:
you can use something as simple as well known Widlar CCS , with additional bjt to other side
voila!
CM , good as Vbe of two current bjt's are equal
I don't think the tail needs to be especially sophisticated. As has been pointed out its balanced at the input and output so the tail is just icing as far as balance is concerned. However I really want to avoid the complexity of a negative rail so a simple current mirror allows this with only 0.7V headroom, which is simply the equivalent of 35R each side of tail resistance.
Shoog
Sorry Zen,
Can not agree though the definition of LTP(aka Schmitt) is close to OT here. Per "common" defintion LTP is SE in differential amp. This one is balanced in.
The CCS´s are for current-balancing. Signalhandling of Shoogs amp will not benefit from them as in a LTP. Actually performance will be bettered with caps parallelling the CCS´s to ground.
But maybe a CCS tail will help if the input transformer is not good enough. The thought is interesting but are there any proofs of this?
Just remember Shoog calling this a "belt and braces"-technique.
It would be better if we could move the currentbalancing up to or "above" the transformer. Rod Coleman have done some work here with currentsensing but I don´t know if it could be applied here.
Can not agree though the definition of LTP(aka Schmitt) is close to OT here. Per "common" defintion LTP is SE in differential amp. This one is balanced in.
The CCS´s are for current-balancing. Signalhandling of Shoogs amp will not benefit from them as in a LTP. Actually performance will be bettered with caps parallelling the CCS´s to ground.
But maybe a CCS tail will help if the input transformer is not good enough. The thought is interesting but are there any proofs of this?
Just remember Shoog calling this a "belt and braces"-technique.
It would be better if we could move the currentbalancing up to or "above" the transformer. Rod Coleman have done some work here with currentsensing but I don´t know if it could be applied here.
Last edited:
I hate to be argumentative here, but there is absolutely nothing in the definition of a LTP which doesn't allow both inputs to be driven. As Zen Mod has said - its all about the tail and nothing about the inputs. To quote Wiki;
"With two inputs and two outputs, this forms a differential amplifier stage (Fig. 2). The two bases (or grids or gates) are inputs which are differentially amplified (subtracted and multiplied) by the pair; they can be fed with a differential (balanced) input signal, or one input could be grounded to form a phase splitter circuit. An amplifier with differential output can drive floating load or another stage with differential input."
The whole principle of operation is to take common signal at the inputs and reject it. In the case of a traditional phase splitter that happens to generate two opposite phase signals of approximately mu/2. In this case it preserves all the mu for gain and forms a composite of the two inputs - hence correcting for any imbalances introduced by an imperfect CTC.
Hope that clarifies.
If it were an option to eliminate the current balancing element then I would simply use unbypassed resistors - but that is not the case and so the cathode to cathode bypass turns it into a medium LTP. Circuits are a series of compromises and these are the decisions which were consciously made along the way. If a circuit can be made simple without significantly compromising performance then that is the design choice I will make. Back to back bypassing has shown itself to work at the cost of raised output impedance - if you fundamentally disagree then it is a simple matter of moving the bypass to ground. In this case that will almost certainly require the 10uf bypass cap be raised - which is another compromise.
Shoog
"With two inputs and two outputs, this forms a differential amplifier stage (Fig. 2). The two bases (or grids or gates) are inputs which are differentially amplified (subtracted and multiplied) by the pair; they can be fed with a differential (balanced) input signal, or one input could be grounded to form a phase splitter circuit. An amplifier with differential output can drive floating load or another stage with differential input."
The whole principle of operation is to take common signal at the inputs and reject it. In the case of a traditional phase splitter that happens to generate two opposite phase signals of approximately mu/2. In this case it preserves all the mu for gain and forms a composite of the two inputs - hence correcting for any imbalances introduced by an imperfect CTC.
Hope that clarifies.
If it were an option to eliminate the current balancing element then I would simply use unbypassed resistors - but that is not the case and so the cathode to cathode bypass turns it into a medium LTP. Circuits are a series of compromises and these are the decisions which were consciously made along the way. If a circuit can be made simple without significantly compromising performance then that is the design choice I will make. Back to back bypassing has shown itself to work at the cost of raised output impedance - if you fundamentally disagree then it is a simple matter of moving the bypass to ground. In this case that will almost certainly require the 10uf bypass cap be raised - which is another compromise.
Shoog
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.