The linear Diode Detector

Ya, a few things,

piglet linked to this detector.

Says it's good to 70Mhz, down 1db. It's uses a high current meter (5ma) but I can work that out.

James Arthur posted a higher frequency high input impedance circuit.

It was suggested I put a voltage divider on the 258 Volt supply before the 24 volt zener, but the tubes warm up slow so, I'm not sure just a series resistor wouldn't work. But, I see the detector has a minimum of 20 ma supply, I don't think the tube regulators are up to that. I'll need to check that.

I'm having trouble with Larkin's posts, text area is blank when I open them.

People want to help but the thread goes off on other topics* rather easily. But, these are often interesting and add to or reinforce my knowledge. * common occurrence

I'm making a parts order for what I need to build Jame's circuit and the detector that piglet posted.

I ask James if I can sub a BF256C for the 2N4416 he specified in his schematic. I think it is OK, but I'm waiting for his reply.

While I'm waiting, I'm off on my own tangent. I thought I'd go through the meter movement error check again. I noticed the meter face does have a linear scale on it. It is 0 to 50 and the 50 aligns with the 250 at FS. I put a 200K 0.01% resistor in series with the meter and adjustable power supply. When I bring the meter up to full scale I get 19.71 volts.

Using 98.55u as FS, I get the following.

Percent of FS Error Percentage

Errors get bigger if I use 100ua as FS.

I'm sending you the spreadsheet, to see if I really messed it up.

I downloaded a faceplate program called Meterbasic from here,

I made a practice faceplate and found to do it right I need to upgrade to the paid version. When the time comes I'll make a face plate that is 100ua FS.

Thanks to everyone for their input, I'll keep building, and refining and hopefully end up with a working circuit. Mikek

Yes, have it correct, I can always adjust the Injection voltage which is factors into the multiplier and keep the needle between the 125 and

Ya, I'm not actually replacing it because of the non linearity as such. The, as built nonlinearity is compensated for by a nonlinear scale on the meter. I'm trying to do two things, one, replace an unobtanium tube. Two, the meter was designed to measure Qs up to 625. I would like to measure Qs up to 2000. That means reducing the injection voltage from 8mV to 2.5mV. When I reduce the injection voltage down to

Thanks, Mike

PS, I sent an email to you Rickman, it bounced while I wrote this.

Don't forget to include the resistance of the meter. You can just measure the voltage across the meter with a volt meter and subtract that from the 5.0 volts and use that to get the current. But this will make the current still lower. Still, a meter, like everything else has tolerance. I don't see 1.5% to be so bad. I bet it was made to be in a

circuit would be 5% tolerance at best. Now we use 1% tolerance resistors as standard. Don't expect so much from the analog meter.

I'm unclear on how the errors get larger with 100 uA full scale. How are you making the measurements?

I'm a bit surprised by that. The measurement circuit (from the tube back to the meter) doesn't know anything about the injected signal level. It simply measures the level across the coil at resonance. The thermocouple measures the injected level. As long as the injected level is high enough to keep the meter reading in the high half the linearity should be the same for all measurements.

Is there something about the thermocouple that can't measure the injected level properly at low voltages? That can be dealt with by calibrating the injection source level against the value of the injected level and measuring the source voltage rather than the voltage across the 0.02 ohm resistor. The injection level can be calculated.

Do you see the same inconsistent measurements for coils with a Q of 500 or lower?

The tube will work the same for all Q values as long as the voltage level is about the same across the coil. The only possible issues with higher Q values is the lower injection voltage may not be measured correctly and the tube circuit could be drawing more current reducing the Q. But that would be opposite with a higher voltage drawing more current and measuring a lower Q I would expect. So I expect the problem is in the thermocouple not measuring the low injection voltage properly.

Did you use the same account as before? I have amdx at knology dot net white listed. I've also set it to not delete spam, but I can't find what the alternative is. Since it is bouncing back to you now, maybe with the delete turned off it will deliver it, but with something in the subject line to indicate spam. Give it another try.

How about using an AD834 multiplier (as a full-wave rectifier) good to ~500MHz? ...Jim Thompson

Well, I didn't actually do the calculation, Maybe I'm wrong. I'll plug it into the spreadsheet and let FS have an error.

Now I'm having trouble with your posts, the one you sent sent at 1:48 my time, has a blank text screen.

Thanks, Mikek

I'm able to see this post just fine. With Thunderbird if a post is downloaded incorrectly it can be corrupted and there is no way to get Thunderbird to delete it and download it again. Stupid Thunderbird.

You need to look into why your IP address is being blacklisted. I did a google search on your IP and found more than one listing service that has reports on your site. This is the one mentioned by the bounce message which I did receive, so your emails are coming through at the moment.

I have seen thermal time constants in the minutes region for separate SMT parts; a dual may be in the second region depending on the internal bonding. If the circuit will allow,use of the same die can get one into the millisecond region.