Best Input differential configuration in amplifiers.

Status
Not open for further replies.
Jump to Latest
#41
LineSource said:
...Matching transistors is a challenge for small quanitity DIY purchases, so I tend to use dual transistors and purchase binned parts. Thermal tracking is also much easier with duals.
Click to expand...

Hi LineSource

For jfet it is very difficult to match but for bjt it is not that difficult ... or maybe I was lucky with my local store. Thermal tracking is interesting though. Do you know some dual bjts that can be used in diff input stage?

Thanks
 
#42
lumanauw said:
Hi, Mike,

Somehow I'm thinking about using very low Hfe like MPSA42 for differential. What bad thing could happen?
Click to expand...

Hi lumanauw, using low hfe-devices for diffamp results in largely
increased inputoffset and very low inputimpedance.
The inputoffset is no problem if you use symetrical ltp.
The low inputimpedance can cause problems like reduced functionality
of feedbackloop, increased distortions, dynamic phaseshifts.
The mainproblem is, that you get a dynamic inputimpedance depending
on the signal flowing inside the ltp. In other words, you get a copy
of the feedbackerrorsignal into the input...
For symetrical ltp you would create 3rd harmonics and for "singleended"
2nd harmonics.
I have very good results with the hfe-monster "mpsa18"...

Mike
 
#44
hi fab
I think using a resistor as a current sourse isn't that bad.
OK, it has times lower CMMR, but active current sources suffer all 'active' imperfections. I mean simple one creates a pole at HF because of base-collector capacitance and cascoded current source is likely to be instable at VHF. All cascode designs may have problems in closed loop because they are complex and add much phase shift at VHF.
regards
 
#45
Thanks everyone very very much for giving valuable comments and discussing the various topologies of differentials.

As for my choice i would go for Symmetric Cascode loaded Fully JFET complementary differential.

Regarding our amplifiers input topology sorry for that , we cannot discuss our design engineering here in the forum.

regards,
Kanwar
 
#46
Workhorse said:
Thanks everyone very very much for giving valuable comments and discussing the various topologies of differentials.

As for my choice i would go for Symmetric Cascode loaded Fully JFET complementary differential.

Regarding our amplifiers input topology sorry for that , we cannot discuss our design engineering here in the forum.

regards,
Kanwar
Click to expand...
So you are collecting ideas from the forum members but you won't discuss your own design.........................?

🙄 🙄 🙄
 
#48
Re: Re: Best Input differential

darkfenriz said:


Hi Elso
the LTP seems normal JFET cascode, but the whole design looks very strange. Folded cascode amplification bases on cerrent through emitter resistors and current mirror forces these currents to be equal... they are bankrupt now, aren't they?😀
Click to expand...


Hi darkfenriz,
What's strange? Or is this way over your head?
The current mirror is presenting a high impedance to the folded cascode stage and provides a differential-input to single-ended-output conversion........
Actually it works ............as a preamp.
:clown:
 
#49
lumanauw said:
Hi, Mike,

Somehow I'm thinking about using very low Hfe like MPSA42 for differential. What bad thing could happen?
Click to expand...
MikeB said:
...The mainproblem is, that you get a dynamic inputimpedance depending
on the signal flowing inside the ltp...
Click to expand...
This is one of the reasons I have become partial to JFET input stages, and also increasingly MOSFET Vas. The extremely high input impedance helps isolate the nonlinearities of the different stages from each other. This is particularly helpful when driving difficult loads.


Elso Kwak said:
How about this one?:
http://www.diyaudio.com/forums/showthread.php?postid=507432#post507432
😎
Click to expand...
That looks unnecessarily complex to me. A JFET cacode, then a BJT cascode in the LTP followed by another (folded) cascode!
 
#50
Mr Evil said:


This is one of the reasons I have become partial to JFET input stages, and also increasingly MOSFET Vas. The extremely high input impedance helps isolate the nonlinearities of the different stages from each other. This is particularly helpful when driving difficult loads.



That looks unnecessarily complex to me. A JFET cacode, then a BJT cascode in the LTP followed by another (folded) cascode!
Click to expand...


Hi Mr Evil,
Ain't that so
Omit the upper CB-pair
guess what, its get rot!
Don't you know the supply is 75 Volts hot!

http://borbelyaudio.com/home_theater.asp
 
#52
Hi, Mr.Evil,

This is one of the reasons I have become partial to JFET input stages, and also increasingly MOSFET Vas. The extremely high input impedance helps isolate the nonlinearities of the different stages from each other. This is particularly helpful when driving difficult loads.
Click to expand...

I always wanted to put MOSFET for VAS. But I have thinkings.
Usually I make differential only 2mA CCS (1mA/leg).
The MOSFET has inner capacitance 600pf-1nF.

For driving 1nF up to 100khz (ideally) it needs much bigger than 1mA, isn't it?

How to drive MOSFET VAS from 1mA differential leg?

I am thinking about TO-220 mosfet like IRF9610, not IRFD
 
#53
lumanauw said:
Hi, Mr.Evil,



I always wanted to put MOSFET for VAS. But I have thinkings.
Usually I make differential only 2mA CCS (1mA/leg).
The MOSFET has inner capacitance 600pf-1nF.

For driving 1nF up to 100khz (ideally) it needs much bigger than 1mA, isn't it?

How to drive MOSFET VAS from 1mA differential leg?

I am thinking about TO-220 mosfet like IRF9610, not IRFD
Click to expand...


Hi LumanauW,

We have used Mosfets in VAS, but there are driven with emitter followers first to get sufficient large bandwidth.

regards,
Kanwar
 
#54
lumanauw said:
Hi, Mr.Evil,



I always wanted to put MOSFET for VAS. But I have thinkings.
Usually I make differential only 2mA CCS (1mA/leg).
The MOSFET has inner capacitance 600pf-1nF.

For driving 1nF up to 100khz (ideally) it needs much bigger than 1mA, isn't it?

How to drive MOSFET VAS from 1mA differential leg?

I am thinking about TO-220 mosfet like IRF9610, not IRFD
Click to expand...
You can get low capacitance MOSFETS which won't be particularly hard to drive. No harm in running the LTP at a high-ish current (5mA total or more) though, or you could use an emitter-follower to buffer it, as Workhorse suggested (also seen in D. Self's designs).
 
#56
lumanauw said:
Hi, Mr.Evil,

How to calculate the bias needed in differential for driving IRF9610 perfectly up to 100khz?
Click to expand...
Looking at the datasheet for that one, it says 170pF input capacitance, which might not present a load to the LTP at high frequencies any worse than a BJT, since that internal capacitance would probably allow you to use a much smaller external Miller compensation capacitor, or even none at all.

But to work out how much current you need, first work out the maximum slew rate you need (differential of the since wave at zero crossing), then the peak current needed to charge the input capacitance I = C dV/dt

Clearly you need at least that much current in each half of the LTP or it will be completely unable to drive the MOSFTET that fast, but for the charging current to have no effect, you need infinite current. Somewhere between those two values is a good compromise😀
 
#58
Hi workhorse!! Have you chosen the best diff input stage yet ?
🙂

And here we go...

Go ahead and use it Kanwar, your amplifiers are going to be superiors and make a breakthrough, with this diff stage you don't need to bother anymore, but don't ask me for permision because I am the wrong person to do so!

Cheers

Ultima Thule has attached this image:
Click the thumbnail to see the original image.
Click to expand...
The document say that Q transistor will increase CMRR . I simulated it but it can't improve any thing and even distortion also increase

I also know which topo is the best . I have just built a amp with current mirror . I found that A normal current is worse than that current mirror . I don't its name
 

Attachments

  • the_perfect_diff_for_workhorse.jpg
    the_perfect_diff_for_workhorse.jpg
    8.9 KB · Views: 475
#59
I also know which topo is the best . I have just built a amp with current mirror . I found that A normal current is worse than that current mirror . I don't know its name
 

Attachments

  • best_current_mirror.jpg
    best_current_mirror.jpg
    3.1 KB · Views: 416
Status
Not open for further replies.
Top Bottom