12at7 LTP CCS

[wicked1]

Hello,
I've been reading about what to set the current to on my CCS for a 12at7 LTP.
Everyone seems to agree on about 3ma, but is that 3ma per side, so set the CCS for 6ma, or is it 3ma between them?

thanks!

[Jeff Yourison]

Hi,

It's 3 mA per side, so 6mA through a common CCS. In Eli Duttman's El Cheapo, which uses a CCS for the 'AT7 LTP, some are running up to 3.5 mA per side. I can't hear a difference, so I keep mine at about 3.2 mA.

-- Jeff

[bigwill]

Regarding this configuration of valves, how balanced is this used as a phase splitter? Given identical anode resistors, is the gain balance near perfect? Is it possible for it to NOT be perfect?

[wicked1]

Thanks, That's what I thought.

Unfortunately I realized it right after I placed my parts order.

[SY]

I'd use a bit less current, more like 2mA. Make the plate loads as large as practical and match 'em- the balance will be as good as your matching, assuming you use a decent CCS.

[kmaier]

The LTP inverter is quite accurate and has good balance, but it's not perfect. Adding a CCS at the bottom will not make it perfect either. As the CCS maintains a constant current flow, whatever one tube increases by, the other tube decreases by. This alone does not result in perfect balance. You can still have imbalance in the pair of triodes used so you should try and source a balanced twin triode (both sections match well). If you want to get the best performance, in addition, I would use a linear pot in the plate circuit as an AC balance control, best adjusted using a distortion analyzer.

Regards, KM

[SY]

You can still have imbalance in the pair of triodes used so you should try and source a balanced twin triode (both sections match well).

With a good CCS as the tail, the triodes should be able to be thoroughly mismatched and the balance will still be as perfect as the loads. Think about it: if the current IS constant (and even in a cheap bipolar cascode CCS, the current is VERY constant), any decrease in one plate current MUST have an equal and opposite response in the other, otherwise Kirchoff's Law is violated. There's no other place for the current to go as long as the stage is not overloaded.

[Salas]

Originally posted by kmaier
The LTP inverter is quite accurate and has good balance, but it's not perfect. Adding a CCS at the bottom will not make it perfect either. As the CCS maintains a constant current flow, whatever one tube increases by, the other tube decreases by. This alone does not result in perfect balance.
Regards, KM

Here is what I use in my amp. Although I could get 5% different DC voltage drops at each plate, for some AT7 tubes in my collection due to differences of gm in the triode pairs, the AC balance was always spot on for AC RMS measurements at the grids of the final stage tubes.

[kmaier]

Yes, I should probably qualify my statement... if you have different gain tubes you can still get some nonlinear performance, not necessarily inversion balance. You might not see it on a scope but a good analyzer will show it. Also, you're probably doing something with the output of the LTP, either driving a pair of output tubes or pre-drivers. This also implies two devices operating 180 degrees out of phase, but it's very unlikely you will have a perfectly matched pair of outputs, as many are only matched to within 5%. As a result, having an AC balance control is a nice option, and adding DC balance to the output stage is also a nice option... you can actually null distortion and output noise to a considerable degree.

Also, the amount of quiescent current in the LTP should be adequate to drive it's intended load at the voltage swing and bandwidth required for your application (ie, slew rate calculation). In the past, I ran a 6FQ7 LTP with a CCS at 10ma. It was driving a pair of KT-88 in UL-PP and required P-P swings in excess of 150 volts. Bandwidth was flat beyond 30KHz.

Regards, KM

[Salas]

Originally posted by kmaier
Yes, I should probably qualify my statement... if you have different gain tubes you can still get some nonlinear performance, not necessarily inversion balance. You might not see it on a scope but a good analyzer will show it. Also, you're probably doing something with the output of the LTP, either driving a pair of output tubes or pre-drivers. This also implies two devices operating 180 degrees out of phase, but it's very unlikely you will have a perfectly matched pair of outputs, as many are only matched to within 5%. As a result, having an AC balance control is a nice option, and adding DC balance to the output stage is also a nice option... you can actually null distortion and output noise to a considerable degree.

Also, the amount of quiescent current in the LTP should be adequate to drive it's intended load at the voltage swing and bandwidth required for your application (ie, slew rate calculation). In the past, I ran a 6FQ7 LTP with a CCS at 10ma. It was driving a pair of KT-88 in UL-PP and required P-P swings in excess of 150 volts. Bandwidth was flat beyond 30KHz.

Regards, KM

Very good additional points. Though it may be argued that some residual harmonic imbalance will be mainly of the second order variety, needed somewhat subjectively in a PP amp to counterbalance the predominating 3rd.
For instance by using that LTP I posted, utilizing the particular non linearities of 12AT7, and not taking measures for absolute harmonic balance of the KT88 output stage more than the actual performance that the PP OPTs could achieve, I got the attached distortion spectrum, where 2nd is stronger than 3rd at -72dB vs -80dB.

(test signal peak is at 100dB, 10dB/div).