Gain Bandwidth Product, or Ft-transition frequency.

[laatsch55]

I know that it's important. WHY is it an important spec, and how is it arrived at??

 

[jbeckva]

That's one of those things I always wondered too, so in looking it up I found this wiki article here. It seems like a good explanation of it for both op amps and transistors... the two are interrelated, obviously.

http://en.wikipedia.org/wiki/Gain–bandwidth_product

 

[laatsch55]

Still kinda fuzzy.

 

[Gepetto]

Lee
GBW is the key determinant in providing flat frequency response in an amplifier. An amp whose frequency response starts to roll off at say 18Khz BECAUSE OF LOW GAIN BANDWITH PRODUCT is suffering because of that parameter.

However an amp with excellent GBW product could also suffer that same early frequency response rolloff if an input filter is set too low.

Less fuzzy?

 

[laatsch55]

No, with all those filters?? How can it bne?? What does the transition frequency have to do with it??

 

[Gepetto]

What filters Lee?

If transistors had infinite GBW product and no Ft effects, then it would be much easier to design ideal amplifiers and tailor the characteristics by applying appropriate filters. Unfortunately DC to daylight amplifiers are not possible. The inherent filter effects of transistors (the Ft parameter) get in the way.

For example, the long tailed pair formed by Q1 and Q2 (take out any of the capacitors in that circuit) has a lot of gain at low frequencies but hits a point at the Ft frequency where its gain tails off at 6dB per octave (single pole). Each succeeding transistor in the amp has its own similar characteristics at different frequencies. These all contribute to a multi pole behavior that has to be compensated for. That is what most of the capacitors in the circuit are attempting to do, kill the gain in a determistic manner before the effects of these transistor poles start contributing.

 

[laatsch55]

Ahhhhhh......................THAT makes sense. So, when researchung for substitutes for any semiconductor in the circuit how far or what is the tolerance for fT? I know,m impossible to answer that one without more specifics, if indeed it is application specific.

 

[Gepetto]

Faster (higher Ft) is usually always better. Most (good) designs do not rely on the Ft for compensation, rather they compensate using other means before the Ft's of the transistors start causing unpredicatable trouble.

For example, in the PL design, C7 (120pF) is a critical compensating component that sets the GBW product of the entire amp.

 

[laatsch55]

Is that a good thing or a bad thing to have one component with that much responsibility?

 

[Gepetto]

Fortunately, 120pF mica caps are very reliable

 

[laatsch55]

So in this case a good thing!!

 

[Gibsonian]

So with the 400 and 700's, current GBP is maxed out?

With the use of the newer transistors in the outputs from original 909's the cap at C7 location is optimized value for largest GBP in these amps? Or is there a bit more involved with the amp circuit design as well?

The GBP explains why output is lower at 20 khz than at say 100 hz, correct?

 

[Gepetto]

I would not say that, it can be further optimized but not by huge amounts. The use of MPSA18 for the Q1 and Q2 positions offers some promise for enhancements that are not achievable with the stock TIS97 devices for example.

 

[Gibsonian]

Was hoping that was the case. Enjoying the ongoing improvements to the amp and don't want it to stop.

 

[laatsch55]

Glad to see you posting again Scott.

 

[Gibsonian]

Still kicking out here in SE Iowa Lee.

 

[laatsch55]

Good to see Bud!!