[Amps] Designing the Cleanest Linear with RF Negative Feedback

[Steve Thompson]

Gary Schafer wrote:
> 
> 
> Steve Thompson wrote:
> 
>>
>>
>> The STC data for 4CX250B gives typical values for class AB1:
>> 2kV anode, 350V screen, 100mA Iq. 2 tone Po 325W pep, IMD3 -30dBc, 
>> IMD5 -50dBc.
>>
>> Same IMD at 1500V anode, 225W pep out.
>>
>> The power levels mean it's not that useful as a guide to what they 
>> will do at much lower level. -30dBc IMD3 is probably somewhere around 
>> 1dB gain compression, and the IMD might be somewhat higher than you 
>> would extrapolate from a compression point prediction. Back the power 
>> off 10dB and you will probably see more than 20dB reduction in the IMD 
>> relative to the tones.
>>
>> If you can find data where someone has used them in a class A TV tx, 
>> you will get better figures for what they will do as 
>> drivers/predrivers in your system.
>>
>> Steve
> 
> 
> 
> Those figures are too good for a 4cx250B. They are probably referencing 
> to pep.

The data sheet is explicit in specifying the levels as relative to each 
tone. In my experience, STC brand 4CX250s do come close to this. There's 
a scanned version at www.g8wrb.org
> 

> The care and feeding book goes on to say that it is not correct to 
> reference distortion levels to pep. If you were to add the levels of the 
> two tone signals together (giving you pep) then you would also need to 
> add the levels of the two 3rd order product together too. This would 
> again give you the same ratio that you get by referencing one of the 3rd 
> order products to one of the two tone levels.

It's ok to reference distortion levels to anything, as long as you say 
what it is.
> 
> They also say that IM product do not necessarily drop as drive is 
> reduced. It can actually be greater at lower power levels than at higher 
> power levels. IM products drop way down at one point midway in the power 
> range and then go back up again. They show a graph that is derived 
> mathematically and say that most tubes follow the graph pretty closely 
> to the predicted IM performance ratios of drive to bias amount.
> 
> It looks like there is around a 10db reduction on average as drive level 
> is reduced to low levels and at some points drops much more than that 
> but only for a narrow range of drive level.

I suspect these characteristics relate to class AB operation - in true 
class A small signal operation, I'd expect the amp to behave with a 
fixed IP3, and IMD to be predictable from that.
> 
> Someone mentioned a few days ago that IM products drop 9 db for a 3 db 
> reduction in power. I believe this is only true when one of the drive 
> signals (one tone of two tones) is dropped in level and the other 
> remains the same. But if both are reduced together and kept at the same 
> level with respect to each other then the 9 db reduction does not come 
> into play.

In the linear region you change the level of both tones by some amount, 
and the absolute level of the IMD changes by 3x that amount. The IMD 
relative to the signal changes by 2x. Change the tones by 3dB, the IMD 
changes by 9dB and the IMD relative to the tones changes by 6dB.

Steve