• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

LTP vs single first stage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
A driver topology much seen recently is an LTP front end driving another LTP. I am wondering what advantages/disadvantages this offers vs. a single tube first stage driving the LTP second stage.

While both topologies have the same overall gain (everything else being equal), a single first stage would put more of the gain up front, which I consider a slight advantage. Also, a single first stage doesn't require a DC balance pot (when the stages are DC-coupled).

Thoughts?

Edit: I'm excluding drivers requiring a balanced input, that one is pretty clear.
 
Yes, I'm assuming fair comparisons, i.e. a single-tube stage would be loaded with a CCS in the plate and LED bias, vs. an LTP loaded with a CCS in the cathode. I'm not seeing superior distortion except perhaps when the PS is not clean and regulated.

A further disadvantage to the LTP is the necessity of a negative supply for the CCS tail.
 
A further disadvantage to the LTP is the necessity of a negative supply for the CCS tail.

This is not necessary. A CCS made with LM317, or with transistors / fets work properly with some 10...20 V voltage drop.

LED bias, which is fixed bias, is not good solution if minimum distortion of the stage is required, since the minimum distortion "sweet spot" can be adjusted especially by fine tuning the cathode resistor.

If you want to have an excellent performing single triode stage, use CCS as anode resistor and adjust the operating point with cathode resistor to minimum distortion.
 
Last edited:
This is not necessary. A CCS made with LM317, or with transistors / fets work properly with some 10...20 V voltage drop.

Yes, I've used LM334, but in an input stage you only have a few volts on the cathode. Anything more than that you have to AC-couple the input, which I would consider a disadvantage.

So aside from distortion, what other advantages are claimed for the LTP?
 
I have taken to always using a LTP at the front, but I also drive both inputs via an input transformer. This allows you to step down the input to allow for better loading of the source. Also it allows you to place the grids at any potential you choose because you reference them through the CT of the transformer to the cathode node of the LTP. In this way you can stack the tail CCS at any point you like. I have used this to place the LTP anodes at zero volts direct coupled to the output stage. This effectively allows a design with no capacitors in the signal path at all.
The Input transformers are not especially demanding and I have pressed many different transformers into service at minimal cost.

Shoog
 
Shoog, I don't follow your logic. Maybe I'm a bit slow today.

If I want that flexibility, I can transformer-couple the input on a simple (common cathode) stage just as well. And shift voltages as desired to get the anode to any DC potential. I don't see how an LTP is inherently superior in that regard. Am I missing something?
 
Shoog, I don't follow your logic. Maybe I'm a bit slow today.

If I want that flexibility, I can transformer-couple the input on a simple (common cathode) stage just as well. And shift voltages as desired to get the anode to any DC potential. I don't see how an LTP is inherently superior in that regard. Am I missing something?

I must apologize for misunderstanding you - you are not talking about using a single triode up front to do your phase splitting - just initial voltage gain. I pictured the input triode as concertina style phase splitting - my mistake.
However if you are going to drive only one input of the LTP, it is my opinion that you are not getting the best out of it performance wise, and so the sooner you have your phase split and are in differential mode the better.

Of course that is just my opinion based on what works for me.

Shoog
 
Last edited:
My experience with LTPs is from the solid state world only. But I don't like them. I don't plan to use them anymore and I much prefer a single device at the input.

Why ?

The LTP works great on paper when you have two perfectly identical devices but when they are different you can have a fair bit of asymmetry. Invariably you need a current mirror to ensure good balance between the currents through each device, and you often need the additional gain that the current mirror provides (see next point) as well as the improved slew rate to drive the compensation capacitor in a feedback amp.

For transistors at least, the transfer curve for an LTP is rather narrow in terms of differential input voltage. It may look enough on paper, but you suffer under transients - the LTP likes small signals. And I've heard the difference - flattening/expanding the transfer curve by using a complimentary feedback pair for the LTP provides remarkable improvements in bass and overall clarity through having a more accurate error amplifier up front. An LTP feeding another LTP is a also a good solution for feedback amplifiers - the additional gain it provides allows more feedback. This reduces the differential voltage at the input LTP so the transfer curve limitations are less severe. But LTPs inside a global feedback look need frequency compensation.

I wouldn't use an LTP as a phase splitter either as I believe a good quality transformer would be inherently more accurate.

Of course single devices have their own problems - dc offset and thermal drift (bipolars).
 
Last edited:
My experience with LTPs is from the solid state world only. But I don't like them. I don't plan to use them anymore and I much prefer a single device at the input.

Why ?

The LTP works great on paper when you have two perfectly identical devices but when they are different you can have a fair bit of asymmetry. Invariably you need a current mirror to ensure good balance between the currents through each device, and you often need the additional gain that the current mirror provides (see next point) as well as the improved slew rate to drive the compensation capacitor in a feedback amp.

For transistors at least, the transfer curve for an LTP is rather narrow in terms of differential input voltage. It may look enough on paper, but you suffer under transients - the LTP likes small signals. And I've heard the difference - flattening/expanding the transfer curve by using a complimentary feedback pair for the LTP provides remarkable improvements in bass and overall clarity through having a more accurate error amplifier up front. An LTP feeding another LTP is a also a good solution for feedback amplifiers - the additional gain it provides allows more feedback. This reduces the differential voltage at the input LTP so the transfer curve limitations are less severe. But LTPs inside a global feedback look need frequency compensation.

I wouldn't use an LTP as a phase splitter either as I believe a good quality transformer would be inherently more accurate.

Of course single devices have their own problems - dc offset and thermal drift (bipolars).
Thats why I always put a transformer in front of the LTP.

Shoog
 
People will turn in their grave at what I am about to say here but I have had success with three types of input transformer;
-OEP microphone input transformer with a 9:1+1 step down ratio for a headphone amplifier
-OEP line input transformer for about €30.00, 2:1+1 step down, a little rolled off beyond about 30khz but adequate.
-mains power toroidals 110V:22V+22V, this is good out to about 60khz but needs loading to stop it ringing out at this frequency - good down to 10hz. This works better than the cheap OEP transformer.

I am currently getting together the parts for a poor mans (cheap as chips) Amnity clone using Talema 115V+115V:15V+15V 7VA power toroidals as both input CT choke and output transformer. I have every reason to expect (from past experience) that this will work perfectly. If using this approach, things to watch out for are, the polarity of the windings as they will ring more easily in one direction, always go for a step down ratio to overcome the inherent interwinding capacitance, and keep all DC out of the windings (Cathode current mirrors will be the used method), and finally go for the smallest VA rating possible to get the most inductance (100H depending on brand). If you expect your transformers to perform flat out to 500khz - then you will have to look elsewhere.

As I said, this is simply to horrific for most people to contemplate, but I have found that I can make a very good sounding product when these details are attended to.

Shoog
 
Last edited:
Morgan Jones covers this in 3rd ed. At the the risk of paraphasing excessively and ruining the message, the balance of a single LTP stage, even if CCS loaded, is significantly lower than a concertina. Multiple CCS loaded LTP stages exhibit comparable balance to a concertina. IIRC this is how the book progresses from the concertina-based Bevois Valley to the 2-LTP-stages-based Crystal Palace.
 
Thats why I always put a transformer in front of the LTP.

Shoog

I must be missing something as I don't see how an input transformer is relevant, you are still relying on the accuracy of an LTP made out of two imperfectly matched devices. A phase splitting transformer of good quality should be more accurate.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.