On 23 Sep 2003 08:31:17 -0700, (Jason Hsu)
wrote:

Believe it or not, I wanted even more QRP capability (like 1mW instead
of 200mW) than my project actually delivers. However, I found several
constraints:
1. Op amps have offset voltages, and these constrain the accuracy of
low-power measurements. Chopper amps have extremely low offset
voltages, but giving them a 10V input while the power supply is turned
off destroys them.
2. The active rectifier has limited accuracy. It's fine in dealing
with 100mV, but it would be useless in rectifying 10mV.
3. The LM3914 chip has limited accuracy. The internal comparator
amplifiers have a few mV of offset voltages.


Hi Jason,

1. Op Amps have offset voltage compensation circuits (either
internal, or you can access 50 years of literature on how to do it
externally). Your choice for a self-destructing chopper amp is your
own problem, not an inherent failure of the class of device. Again,
there are 50 years of literature on how to build your own if the
devices your limit yourself to are either a. too expensive, or b. too
fragile.

2. Active rectifiers having limited accuracy is strictly a problem of
GBWP. Choose a crappy one, and you fulfill your nightmare. For HF,
you should be using one with at least 1GHz. Your problem then becomes
one of selecting an amp that is stable at unity gain (or at least
stable at the gain you choose/need). Complaints about not being able
to handle 10mV signals only suggest you need an amplifying buffer
before the detector. With a little leg work (researching that same 50
years worth of application design), a second diode, and feed back, you
could linearize the power/SWR meter too.

3. LM3914 should be designed with the usual offset compensation if
that is a problem. However, even with a few mV out of 10V, the
dynamic range is considerably greater than your instrument's range.
Your problem is one of scaling your signal (it shows in the other
complaints). Dynamic range, 20 log (10V/10mV), borders on 60dB where
accuracy would tend to go to hell at the low end. You show only 30dB
of dynamic range, 10 log (200W/200mW). You are not using the full
potential of the device.

Compensation circuits can run from simple to complex. If you have
zero compensation, you are abandoning your fate. Reading National
Semiconductor product literature would be a simple approach to a great
deal of variety of design:
http://cache.national.com/ds/LM/LM3914.pdf

73's
Richard Clark, KB7QHC