|
Vintage Test Gear and Workshop Equipment For discussions about vintage test gear and workshop equipment such as coil winders. |
|
Thread Tools |
2nd Mar 2019, 3:40 pm | #61 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
I think you could use the shunt 'insertion loss' method with your sig gen if you built a basic log amp detector using an AD8307 chip. This would be a cheap alternative.
However, to define the 50R test port accurately you would need to use at least a 10dB attenuator on the sig gen side and I'd recommend an attenuator on the AD8307 (detector) side as well. As Hayward mentions, the accuracy of the port impedance of the detector is less important with this method but I would still use an attenuator here too. The disadvantage with this logamp method is that it will be harder to find the null because you can't see a swept display. Something like the Analog Discovery 2 plus a suitable jig would also work well here I think because it would provide a swept display and excellent log accuracy. The log accuracy of the AD8307 is very good but not in the same class as the AD2 with its 100MSa/s 14 bit ADC. I have an old 'power meter' I made over 20 years ago that uses the AD8307. It was made when the AD8307 was first released and I recall that the log accuracy is quite good over a range of >60dB. I put a decent attenuator and compensation network inside it to get a very low port VSWR. Otherwise, if you stick with an E/e or a -3dB Q measuring system that needs a high impedance detector that connects directly to the test circuit then you will always be battling against the limitations of the detector. HP obviously managed to make a high Z detector that behaves up to 70MHz but their circuit is quite complicated. The spec for the Marconi TF1245 detector shunt resistance is 12Meg at 1MHz. This drops to 0.3Meg by 100MHz. They achieve this with a valve (diode) detector and presumably they manage it without any negative resistance issues anywhere. As others have said already, you will probably be fine with an Advance T2 Q meter if you only want to look at LW and MW stuff with moderate Q. It will probably give respectable performance at higher frequencies too. The homebrew meters shown on this thread would probably give reasonable results when used within these limits. However, try and measure higher Q coils at higher frequencies and they will all have significant issues I think.
__________________
Regards, Jeremy G0HZU |
2nd Mar 2019, 4:59 pm | #62 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,924
|
Re: Q-meter questions
Although this thread is about a Q-Meter, the problem which is cropping up is actually with the high-impedance voltmeter which needs to be at the back end of it. Consequently, we are covering the same ground as we did in Skywave’s thread back in 2016, and other threads about RF probes.
Let me put my head above the parapet and ask the question, what if we forget about FET’s and think Vintage Radio… a valve, possibly a Nuvistor? Could we steal any ideas from Tek’s oscilloscope designs which used Nuvistors? I did spend some time playing with Nuvistors back in 2016, but I just don’t have the expertise to design circuits so didn’t get a result, but I’m sure someone else might well do better. B
__________________
Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch. |
2nd Mar 2019, 7:17 pm | #63 | |
Heptode
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
|
Re: Q-meter questions
Quote:
There are lots of advanced technical materials in this thread. I think it will take me a while for me to digest and "sink in". With regard to log amp detector I think you are referring to something like this: http://www.crystal-radio.eu/enqmeting.htm His high impedance, low capacitance FET amp schematic: http://www.crystal-radio.eu/fetamp/enfetamp.htm I don't have a degree in electronics or know much about modern electronics. I am still reading vintage valve books published 80 years ago. But I am familiar with the Q measurement using the ring-down method which is the same technique that I often use during the flight tests of the damped oscillation of an aircraft (my background is aerospace engineering). This is the ring-down method with a 10M coupling resistor and 10:1 probe: https://www.youtube.com/watch?v=sJ3mRAVo5n0 In this test(video), the voltage sensitivity of the scope is important. With this method, both 10:1 and 100:1 probes won't work with low-end basic scopes having poor vertical sensitivity mV per division. I have tried the above ring-down method with my West German Hameg HM203-6 (given to me free), it simply does not have enough vertical sensitivity. This is the cheap Ukrainian active probe I am talking about (I dont' care about care how rubbish it is at 1GHz. It is good enough at MW frequency): https://www.eevblog.com/forum/rf-mic...obe-from-ebay/ My new, cheap Picoscope has died and I will have to return it. I am toying with the idea of getting a Rigol DS1054Z or Siglent SDS1202X-E 200MHz, but I am fighting the urge Last edited by regenfreak; 2nd Mar 2019 at 7:37 pm. |
|
2nd Mar 2019, 7:30 pm | #64 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: Q-meter questions
For those frequencies there are plenty of valves which can offer high impedances. The nuvistorishness of nuvistors doesn't really offer any electrical advantages. Tek and HP used them because of their small size.
However, the equations Jeremy's been using relating to the creation of negative resistance and instability for FETs also apply to bipolar transistors and to valves. With thermionics, you run into the same tradeoffs. The HP FET Q-meter in order to follow the magnified voltage in the tank used higher than usual supply rails of + and - 25v to allow the FET to follow the tank voltage. Fancy bootstrapping was used to keep Z high even at the upper end of the frequency range (and this was only 70MHz) The higher voltages inherent in valve circuitry will help in this respect. But heater/cathode C is a limitation transistors don't have. Many organisations have developed a lot of radio stuff and never owned a Q meter. You can measure components with vector network analysers, with Rx meters, with vintage GR RF bridges (Wayne Kerr did one too) but these start getting very inaccurate at measuring the losses of high Q components. And so do Q meters! but the Q meters are useful a bit higher in Q. Quartz crystals have Qs in the 10,000 to 1 million region and frequencies to well over 100MHz. So based on the difficulties with a simple Q meter, then measuring these must be impossible...... so there's a puzzle David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
3rd Mar 2019, 3:01 am | #65 |
Heptode
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
|
Re: Q-meter questions
Here is the article that I was referring to..by Jacques Audet about the transmission method using the 1pF coupling caps, shunt mode transmission and reflection transmission SWR analyser method :
http://hb9abx.no-ip.biz/VE2AZX-Q-factor.pdf Here is the "Pennywise" active probe that is the same design as the the Ukrainian one: https://elektrotanya.com/files/forum...10/e04a036.pdf |
3rd Mar 2019, 5:21 am | #66 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,924
|
Re: Q-meter questions
Given the admiration that's being aimed at the Advance T2, I was pleased to find that this was discussed in some detail and the full manual for it published on our very own forum, post 9 here https://www.vintage-radio.net/forum/...ad.php?t=70138. Looks like it could be copied?
B
__________________
Saturn V had 6 million pounds of fuel. It would take thirty thousand strong men to lift it an inch. |
3rd Mar 2019, 8:04 am | #67 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: Q-meter questions
There are a few awkward areas in it. The tapping on the mains transformer HT isn't at a disclosed voltage, but the pot following it compensates.
Note that this instrument amplitude modulates its oscillator, then detects and amplifies the modulation found at different places in the test tank. It might be a bit susceptible to the distorted mains waveforms we get today, but setting the oscillator reading should compensate. The big uncertainty is duplicating the low inductance bar and that loop inductor. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
3rd Mar 2019, 2:35 pm | #68 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
Quote:
As long as the sig gen has low harmonics (-50dBc would be good) and the test ports have attenuators on them to define an accurate 50 ohms then this method should be very accurate. At resonance, the LC network reduces to just a resistor so this method becomes a potential divider. If the AD8307 has good log accuracy then you should be able to measure Q at least as well as any other method mentioned so far. You could check the accuracy using a series of 1% metal film resistors in place of the LC network. You should get very pleasing results The only method that can (usually) compete is my other method that measure the 3dB bandwidth using E and H field probes but this is fiddly to use and is best done with a VNA. I also semi automate this method using some software I wrote to control the VNA and it measures the 3dB bandwidth. With this method there is no source impedance error and no load impedance error. The ESR of the capacitor needs to be low and I usually use 3 ATC caps in parallel up at VHF.
__________________
Regards, Jeremy G0HZU |
|
3rd Mar 2019, 2:55 pm | #69 | |
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Q-meter wanted
Quote:
This online calculator from M0UKD is excellent, and measurement of the as-built just confirms that the coil is fit for purpose - the labour is already done and is irreversible (so you'd have to scrap one that was no good and start over again). You'll usually converge on a Q of between 400-500. Interesting thread,though!
__________________
Al |
|
3rd Mar 2019, 3:06 pm | #70 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
Quote:
I must confess that I was too lazy and tired last night to count the number of cycles to the point where the level drops by half. But I estimated it using the X axis scale because the ringing frequency is about 547kHz. So I think the method indicates that the Q is about 330 at 547kHz. The inductor is a 50uH coil wound on an old Siemens pot core. It's a very small coil, about the size of a sugar lump. I resonated it with three 560pF Suflex caps in parallel.
__________________
Regards, Jeremy G0HZU |
|
3rd Mar 2019, 3:54 pm | #71 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
I measured the same coil and capacitor combo using the classic series LC in shunt method on my VNA. It predicts the Q is 376 at 547kHz. There is no source impedance error to correct for here because the jig has a very accurate 50R source and load impedance. There is the ESR of the caps to consider but this affects both methods the same.
If I use the Rowe formula to correct for the 0.072R source impedance then the ringdown Q of 330 corrects to 382.5. So this is very close to the Q of 376 I got with the VNA. However, I didn't measure the 50% point very accurately with the ringdown method and I'm not sure how accurate my 0.072R source is at 547kHz. So there is a lot of uncertainty here even though the numbers agree quite well once corrected.
__________________
Regards, Jeremy G0HZU |
3rd Mar 2019, 4:26 pm | #72 | |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: Q-meter questions
Quote:
Where the ATC caps win is in robustness. They survive things that would flash red lesser components. I stick 300W GHz bursts throuch ATC capacitors. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
|
3rd Mar 2019, 4:53 pm | #73 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
Yes, the 100 series aren't that great. I generally use either the ATC 600S series or the 800B series for stuff like this although we do sometimes use the 100B and 600F series at work.
At work we recently received a bumper sample kit of PPI caps and these seem to be very good as well. They sent us a shoebox full of sample bags. 10 of each value including lots of 'in between' values. This was across several series of caps so there was probably over 1500 caps in the sample box. Mainly the equivalent of 100 series ATC and also 0805 and 0603 and 0402 series. I've used these a lot for casual dev stuff at work and they seem to be close to the ATC equivalents. http://www.passiveplus.com/kits.php
__________________
Regards, Jeremy G0HZU |
3rd Mar 2019, 5:04 pm | #74 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
I tried a quick lashup of the AD8307 method using the same inductor and cap combination. I used precision 10dB attenuators on the jig (easily made using 1% resistors) and this improves the accuracy of the 50R port impedance.
I used a sig gen that had harmonics at -45dBc or better so I didn't need a LPF to remove these. I used my 20 year old AD8307 based power sensor as the detector. I've attached an image of a quick and dirty excel sheet to show how I calculated the Q based on the test frequency that gives a null, the test capacitance, the Jig Z of 50R, the gain slope of the AD8307 (0.025V/dB as per the datasheet) and the AD8307 RSSI voltage readings for through calibration and also the RSSI voltage when the LC resonator is added. It gives good results as you can see in the excel sheet image below. It predicts the Q is just over 380. The datasheet spec for the gain slope is 0.025V/dB +/- 0.002 but you could prove this using several precision (or even just 1%) shunt resistors in the jig to see if the AD8307 reports this correctly. But for most people measuring inductors with moderate Q this method should be fine without doing this. However, there will be some coils that have an ESR up towards 50 ohms or more and the jig won't be as accurate here. But you could make a jig with a higher port impedance or just accept a bit more uncertainty in the result.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 3rd Mar 2019 at 5:10 pm. |
3rd Mar 2019, 5:42 pm | #75 | ||
Heptode
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
|
Re: Q-meter questions
Awesome Jeremy
Quote:
https://www.digikey.co.uk/en/article...loscope-probes I have measured the DC resistor of my generic Chinese 10:1 probe set at 1:1, it is 100 Ohms. This is what I call the "Black Magic" of transmission line optimised design. Quote:
The theory of log Amp AD8307 is very involved.I would probably stay away from it. I have come across no less than 10 methods of measuring Qu. For me the ring-down method is the easiest. If you are a Ham guy and own MFJ-259C SWR antenna analyser, the Reflection measurement method in Jacques Audet's article is the way to go. |
||
3rd Mar 2019, 5:51 pm | #76 | ||
Heptode
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
|
Re: Q-meter questions
Quote:
Quote:
|
||
3rd Mar 2019, 7:32 pm | #77 |
Heptode
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
|
Re: Q-meter questions
I know nothing about transmission line theory as it is like antenna theory; they are like witchcraft and voodoo, on page 5, the characteristic impedance of a non-ideal coaxial cable is given. The signal behaves like wave, scary stuff! The skin effect resistance is proportional to square root of angular frequency:
http://pmaweb.caltech.edu/~ph77/hand..._11-2-2017.pdf |
3rd Mar 2019, 7:38 pm | #78 | ||
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
Quote:
So the 0.072R impedance of my test source is something like 15% of this. It adds to the ESR of the 50uH inductor and spoils its Q. So if I don't correct my measured Q (after doing the ring down method with the 0.072R source) then I'll get a result that is a bit lower than it should be. Quote:
A similar damping error obviously occurs if you damp the LC circuit using probes as the probes will place a shunt resistance across the circuit under test and this spoils the Q. However, you would need a revised version of the equation to cope with shunt damping as the Rowe equation is for series damping.
__________________
Regards, Jeremy G0HZU |
||
3rd Mar 2019, 9:10 pm | #79 |
Heptode
Join Date: Nov 2018
Location: London SW16, UK.
Posts: 655
|
Re: Q-meter questions
Thanks Jeremy
I will go away and think about it This explains the mystery of why scope cable has the dc resistance of 50 to 200 ohms: https://www.dfad.com.au/links/THE%20...ES%20OCt09.pdf I must say it is a very good article Last edited by regenfreak; 3rd Mar 2019 at 9:32 pm. |
3rd Mar 2019, 11:41 pm | #80 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: Q-meter questions
Yes, a typical x10 scope probe will often prove to be unsuitable as a high impedance sniffer.
At RF frequencies, the equivalent shunt resistance of a x10 scope probe will drop quite quickly. By 5 or 6MHz it could be as low as 100,000 ohms in parallel with about 12pF. By 20MHz that will drop to maybe 8,000 ohms in parallel with 12pF. So this is not good at all and will usually damp a tuned circuit. For example, a 1uH inductor wound on a T50-6 powdered iron toroid might have a Qu of about 200 at 20MHz and this would have an Rp of 25,000 ohms. So that is about three times the Rp of a x10 scope probe at the same frequency! So the x10 scope probe would damp the Q of the inductor from 200 to about 50 at 20MHz.
__________________
Regards, Jeremy G0HZU |