Jan, just compare my buffer with the one Walter Jung has drawn and to a regular one, BUF634 for instance. The difference is cascodes, transistors which only act as diodes and R14 and R15. Why have I done this? Only to limit the amount of different parts. How would you done a diamond buffer?
The pullup, pulldown resistors is just in case only. You can use this buffer with single supply.
All measurements is without feedback, just the buffer alone. I have though a reservation about the measurements but Walt Jung claimed rather good performance and when was he totally wrong?
SMD is "ont i stjärten"(*) when it comes to repair but don't break it.... just design in protection which works if it is needed. In my case I have no protection because it should work in known conditions. Walter Jung had suggestions how it could be done.
The loads I have used are 100k, 220 ohms and 55 ohms.
(*) Fred will translate for you...
The pullup, pulldown resistors is just in case only. You can use this buffer with single supply.
All measurements is without feedback, just the buffer alone. I have though a reservation about the measurements but Walt Jung claimed rather good performance and when was he totally wrong?
SMD is "ont i stjärten"(*) when it comes to repair but don't break it.... just design in protection which works if it is needed. In my case I have no protection because it should work in known conditions. Walter Jung had suggestions how it could be done.
The loads I have used are 100k, 220 ohms and 55 ohms.
(*) Fred will translate for you...
I felt like it but it's not the Fred style exactly. It's super universal, inverting och non-inverting buffer, inverting or non-inverting LM3886, DC-servo or not for both inverting and non-inverting LM3386, bridge connection in inverting or non-inverting mode with non-inverting or inverting input buffer with or without input coupling caps along with an optional output filter plus an input filter....have I forgot something?... it's not SMD :PMA said:
:up
What is the Walt Jung article you are referring to, Per? Is it available on the web?
Regards,
Eric
Regards,
Eric
http://www.elecdesign.com/Globals/PlanetEE/Content/2800.html
The rest is gone for the moment, can't be found. Maybe they will turn up some day.
Scanned articles at Walt's homepage.
http://home.comcast.net/~walt-jung/wsb
The rest is gone for the moment, can't be found. Maybe they will turn up some day.
Scanned articles at Walt's homepage.
http://home.comcast.net/~walt-jung/wsb
I have done a simulation of this diamond buffer and I get a bandwidth of around 300 MHz. The BC860, 850 were changed to 2N3904 and 2N2907 but the output transistors were the right ones, BCP53 and BCP56.
Is a bandwidth clearly over 100 MHz possible?
Is a bandwidth clearly over 100 MHz possible?
NO!, are you sure about this sim. Try with another spicemodel.
If i remember right i have gotten something like 10 - 20MHz.
If i remember right i have gotten something like 10 - 20MHz.
Hi Per.
Yes you are right about the wide bandwidth ...
I got 260MHz with a source impedance of 1 Ohm.
I got 123MHz with a source impedance of 100 Ohm.
And i got 35MHz with a source impedance of 1 KOhm.
But i would say that in a real world application The buffer will run around 1KOhm - 20Ohm at > 20MHz.
A fast opamp like the AD8610 will have a Zout peaking at 60MHz of 50Ohm.
Yes you are right about the wide bandwidth ...
I got 260MHz with a source impedance of 1 Ohm.
I got 123MHz with a source impedance of 100 Ohm.
And i got 35MHz with a source impedance of 1 KOhm.
But i would say that in a real world application The buffer will run around 1KOhm - 20Ohm at > 20MHz.
A fast opamp like the AD8610 will have a Zout peaking at 60MHz of 50Ohm.
I know that the source impedance has strong influence of the bandwidth.
Now I'm trying to get FFT which wasn't easy to remember how it was done, but thanks to andy_c I can read what he says about it.
Now I'm trying to get FFT which wasn't easy to remember how it was done, but thanks to andy_c I can read what he says about it.
Hi Per.
Check the Hfe curve for the Spice model before the FFT test. You will be surpriced how bad some of them are.
Just a hint.
Check the Hfe curve for the Spice model before the FFT test. You will be surpriced how bad some of them are.
Just a hint.
Source impedance doesn’t have an influence on the bandwidth. It only forms the low pass filter with the input capacitance.peranders said:
Pedja
Yeah? Put in 100 ohms (or whatever) in series with the signal source and you will get reduced bandwidth.Pedja said:
You will get a low passed circuit. The circuit's bandwidth remains the same.peranders said:
Pedja
I have run at 16384 steps. Oki, doki. I have increased to 65536, better.....Pedja said:
What about the dynamics. The FFT shows only around 100 dB S/N. Has this something to do with LTSpice?
You may also need a longer sample train and/or shorter step times to get really nice looking FFT. I usually play with all three settings when I am trying to find low-level behavior.
In tools>control_panel
Remove all compression.
Run transient with the smallest timestep like 2msec window, 20nsec step.
Run it for at least 20 periodes.
1KHz => 20msec, 20nsec
20KHz => 1msec, 10n
10KHz => 2msec, 20n
Remove all compression.
Run transient with the smallest timestep like 2msec window, 20nsec step.
Run it for at least 20 periodes.
1KHz => 20msec, 20nsec
20KHz => 1msec, 10n
10KHz => 2msec, 20n
- Status
- Not open for further replies.