Hi Dave:
One of the key things about the Kobe Instrument Division (KID) is
that they pay a lot of attention to the errors surrounding the
measurements their instruments make.? An excellent reference is The
Impedance Handbook.? There may be a computer program (maybe an app)
that will make those calculations.
--
Have Fun,
Brooke Clarke
axioms:
1. The extent to which you can fix or improve something will be limited by how well you understand how it works.
2. Everybody, with no exceptions, holds false beliefs.
-------- Original Message --------
On Wed, 10 Aug 2022 at 18:59, Mikek < amdx@...>
wrote:
Doc, that's great, most
of use are not about to spend $14,000 on three pieces of
equipment for our hobby needs.
Especially when $80 will get us close enough and allow us to
compare coils.
?I'm on other groups where that would be called a humble
brag.
????????????????????????????????????????? Mikeka
Mikeka,
if you look at the description of the group
it says "There are no limits to the complexity or
cost of the design, so both a cesium fountain clock and a
basic voltmeter are acceptable." So the group is
not just for hobbies.?
The total outlay for that equipment, which was all bought
on different times and in different places, was around
$5,000 not $14,000. That equipment was bought for a business
and used by a business. But none of it is good at measuring
Q. ???
So for me, that Chinese kit does not do what I want, but
neither does anything my company owns do what I want.???
Hence my desire to design and make something that works
better than the instruments I have.?
Dave
|
On Wed, 10 Aug 2022 at 18:59, Mikek < amdx@...> wrote: Doc, that's great, most of use are not about to spend $14,000 on three pieces of equipment for our hobby needs.
Especially when $80 will get us close enough and allow us to compare coils.
?I'm on other groups where that would be called a humble brag.
????????????????????????????????????????? Mikeka
Mikeka,
if you look at the description of the group
it says "There are no limits to the complexity or cost of the design, so both a
cesium fountain clock and a basic voltmeter are acceptable." So the group is not just for hobbies.?
The total outlay for that equipment, which was all bought on different times and in different places, was around $5,000 not $14,000.
That equipment was bought for a business and used by a business. But none of it is good at measuring Q.
???
So for me, that Chinese kit does not do what I want, but neither does anything my company owns do what I want.??? Hence my desire to design and make something that works better than the instruments I have.?
Dave
|
Doc, that's great, most of use are not about to spend $14,000 on three pieces of equipment for our hobby needs.
Especially when $80 will get us close enough and allow us to compare coils.
?I'm on other groups where that would be called a humble brag.
????????????????????????????????????????? Mikek
|
On Wed, 10 Aug 2022 at 18:10, Mikek < amdx@...> wrote: If anyone has an interest in the updated version of the Ring Down Q Meter KIT,? they can contact? kwanmingmak@...? and put in a preorder.
They have not been built in great quantity yet, but there is a an order to build a small quantity. Expected in 2 months.
It measure up to a 10Mhz and down to 10uH. It is the unit shown here, I have found Google Translate does very good on some posts and not as good on other posts.
For me at least, it does not offer much other than interest. I have
HP 4284A Precision LCR meter - 20 Hz to 1 MHz HP 4285A Precision LCR meter 75 kHz to 30 MHz HP 4291B Impedance analyzer 1 MHz to 1.8 GHz
The frequency range and inductance range of that kit is too small. But I note it can read Q's up to 2500. I wonder what the uncertainty is? How do you verify it?
The ring down method is certainly an interesting way of measuring Q, but you still have the limitation of the more common method, in that the detector circuit loads the coil. I don't think there's any way to square that circle.
Dave
|
If anyone has an interest in the updated version of the Ring Down Q Meter KIT,? they can contact? kwanmingmak@...? and put in a preorder.
They have not been built in great quantity yet, but there is a an order to build a small quantity. Expected in 2 months.
It measure up to a 10Mhz and down to 10uH. It is the unit shown here, I have found Google Translate does very good on some posts and not as good on other posts. http://www.crystalradio.cn/forum.php?mod=viewthread&tid=1613531&page=11#pid23237736
|
Yes, in an email yesterday, Ming told me that 20 units will be produced,
Ming said it will be 2 months. These are just the boards to assemble a unit.
????????????????????????????????????????????? Mikek
|
I had contact with the distributor for these kits..
It seems that a new kit will be available soon, extension to 10 Mhz
Perhaps interesting
JP
|
Are you familiar with the high Q fixed value capacitors that were recommended for use with that Ring Down Q meter?
I think they were characterized for Q. If you know the Q of a cap, you can subtract it out of the total LC Q and know the
true coil Q. From there you could characterize other capacitors with a known true Q. Unless you have a Q over frequency graph,
you may still be stuck at only knowing Q at one frequency.
|
Hello
Kits and PCB's are available for sale for this last meter.. ( Ring Down Method )
But it works only to +/- 2 Mhz, and new design is made though by the same Chinese guy, but no PCB's exist, for the new version
For HF purposes, we need more..
JP
|
There's a YouTube video, "Demo of a DIY digital L/C Circuit Q-Meter" <>
where the author, demonstrating a Q meter, tests an inductor using a number of different variable capacitors. Wasn't a rigorous or well controlled test, but a considerable difference in Q with different caps.
John
toggle quoted message
Show quoted text
On 8/9/2022 6:04 AM, Dr. David Kirkby, Kirkby Microwave Ltd wrote: On Tue, 9 Aug 2022 at 13:02, Steven Greenfield AE7HD via groups.io <> <alienrelics@... <mailto:[email protected]>> wrote:
I don't think I've seen anyone mention the impact of the variable
capacitor on Q. I understand that for lower Q, the losses in the
inductor swamp losses in the capacitor. But at what point do
capacitor losses begin to impact the measurements?
In other words, will the garden variety tin coated, brass, or
aluminum plate variable do? Does it need to be silver plated for
lowest losses?
Steve Greenfield AE7HD
Somewhere I read that the capacitor loss can become significant above 10 MHz. So cheap capacitors would be yet another source of error. I would guess that a butterfly capacitor, where there’s no sliding contact would be preferable. Likewise silver plating.
I can’t help feeling that the answer to the losses *might* be to introduce a further known loss, to determine the original unknown loss.
Take for example an analog multimeter (eg AVO). If that is measuring a voltage from an unknown source impedance, and the impedance of the multimeter is unknown, you don’t know the source voltage, as there are two errors. I would have thought that using a couple of known resistors, one could deduce the internal resistance of the multimeter and the source resistance, allowing the source voltage to be determined.
Dave
--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@... <mailto:drkirkby@...>
<>
Telephone 01621-680100./ +44 1621 680100
Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom
|
I'd suggest using a butterfly type capacitor with an isolated rotor and no sliding contacts, or perhaps a small bellows type vacuum cap if you are chasing the best possible Q.
Somewhere I have a really nice small 1500 pF ceramic vacuum cap.? I'll find it one day... Also definitely a good idea to enclose the cap it is isn't a vacuum type. These days I usually flush sealed enclosures with Argon from my welding set before sealing them, and fit a Gore-Tex or similar breather filter to limit insect and dust ingress and avoid diurnal pumping.
Neil G4DBN
toggle quoted message
Show quoted text
On 09/08/2022 15:07, Pete_G4GJL wrote: Some Q-Meter variable capacitor design thoughts, in no particular order:
Variable capacitor?design should incorporate wide and large contact surface area? brushes for rotor connection (s). These should be cleaned chemically and coated with Caig DeOxit or similar?compound to maintain long life and low resistance connection. Do not mechanically burnish the moving contact surfaces: if they are dirty use several rotations of the moving vanes to re-establish a good (maximal) contact area. Watch as the bright surface re-appears and then add the DeOxit.
Use a wide spacing variable capacitor with silver plated smooth plates. The designer of the HP Q-meter polished the plates of the variable capacitor in the HP4342.
Ceramic, quartz or glass insulation is far better than phenolic or plastic materials.
Broad, tightly fitting ground connection to chassis is required.
Broad 'HI' or stator connection is required. Silver plated copper ribbon, as short as possible, can be used. Place the stator connection to the variable capacitor as close to the test terminals as possible to minimise this distance.
Maintain a dust cover (or screening) around the moving parts.
|
OOPS! My post overlapped some ideas offered by you and others, PeterS
Peter Worrall G4GJL
Steve,
Yes and maybe.? You need to consider the application and expected Q to be measured and do some calculations to see if it is important to you.? If Q is less than a couple of hundred, the losses in an air-spaced variable capacitor are probably too small to worry about unless it is old (most are), has signs of corrosion, and/or the grease for the rotor bearings/contact fingering has dried out.? Most Q meters have special? arrangements to minimise contact resistance and keep it stable.
If you think you have to worry about that, you also need to consider the rest of the circuit between the inductor connections and the capacitor.? If you look at photographs in the HP Journal for the HP4342A or the manual for the Marconi TF1245A you will see that the connections are substantial silver-plated brass.? But for a home brew Q meter to do A-B comparisons rather than absolute measurements, a well made air-spaced variable capacitor in good or restored condition and thick wire or braid/brass sheet will do.
Wes Hayward, W7ZOI , carried out many careful measurements on inductors and air spaced variable capacitors and found most of the capacitors had a Q of around 350 at 1MHz with his "best" toroid.? (See "Experiments with Coils and Q measurements" amongst several articles he published).
PeterS??? ??? G8EZE
-----Original Message-----
From: Steven Greenfield AE7HD via <alienrelics= [email protected]>
To: [email protected]
Sent: Tue, 9 Aug 2022 13:02
Subject: Re: [Test Equipment Design & Construction] Making a Q-meter /
I don't think I've seen anyone mention the impact of the variable capacitor on Q. I understand that for lower Q, the losses in the inductor swamp losses in the capacitor. But at what point do capacitor losses begin to impact the measurements? In other words, will the garden variety tin coated, brass, or aluminum plate variable do? Does it need to be silver plated for lowest losses? Steve Greenfield AE7HD
|
Some Q-Meter variable capacitor design thoughts, in no particular order:
Variable capacitor?design should incorporate wide and large contact surface area? brushes for rotor connection (s). These should be cleaned chemically and coated with Caig DeOxit or similar?compound to maintain long life and low resistance connection. Do not mechanically burnish the moving contact surfaces: if they are dirty use several rotations of the moving vanes to re-establish a good (maximal) contact area. Watch as the bright surface re-appears and then add the DeOxit.
Use a wide spacing variable capacitor with silver plated smooth plates. The designer of the HP Q-meter polished the plates of the variable capacitor in the HP4342.
Ceramic, quartz or glass insulation is far better than phenolic or plastic materials.
Broad, tightly fitting ground connection to chassis is required.
Broad 'HI' or stator connection is required. Silver plated copper ribbon, as short as possible, can be used. Place the stator connection to the variable capacitor as close to the test terminals as possible to minimise this distance.
Maintain a dust cover (or screening) around the moving parts.
Peter Worrall G4GJL
On Tue, Aug 9, 2022 at 2:05 PM Dr. David Kirkby, Kirkby Microwave Ltd < drkirkby@...> wrote: On Tue, 9 Aug 2022 at 13:02, Steven Greenfield AE7HD via <alienrelics= [email protected]> wrote: I don't think I've seen anyone mention the impact of the variable capacitor on Q. I understand that for lower Q, the losses in the inductor swamp losses in the capacitor. But at what point do capacitor losses begin to impact the measurements?
In other words, will the garden variety tin coated, brass, or aluminum plate variable do? Does it need to be silver plated for lowest losses?
Steve Greenfield AE7HD
Somewhere I read that the capacitor loss can become significant above 10 MHz. So cheap capacitors would be yet another source of error. I would guess that a butterfly capacitor, where there’s no sliding contact would be preferable. Likewise silver plating.?
I can’t help feeling that the answer to the losses might be to introduce a further known loss, to determine the original unknown loss.?
Take for example an analog multimeter (eg AVO). If that is measuring a voltage from an unknown source impedance, and the impedance of the multimeter is unknown, you don’t know the source voltage, as there are two errors. I would have thought that using a couple of known resistors, one could deduce the internal resistance of the multimeter and the source resistance, allowing the source voltage to be determined.?
Dave?
Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@...Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom
|
Steve,
Yes and maybe.? You need to consider the application and expected Q to be measured and do some calculations to see if it is important to you.? If Q is less than a couple of hundred, the losses in an air-spaced variable capacitor are probably too small to worry about unless it is old (most are), has signs of corrosion, and/or the grease for the rotor bearings/contact fingering has dried out.? Most Q meters have special? arrangements to minimise contact resistance and keep it stable.
If you think you have to worry about that, you also need to consider the rest of the circuit between the inductor connections and the capacitor.? If you look at photographs in the HP Journal for the HP4342A or the manual for the Marconi TF1245A you will see that the connections are substantial silver-plated brass.? But for a home brew Q meter to do A-B comparisons rather than absolute measurements, a well made air-spaced variable capacitor in good or restored condition and thick wire or braid/brass sheet will do.
Wes Hayward, W7ZOI , carried out many careful measurements on inductors and air spaced variable capacitors and found most of the capacitors had a Q of around 350 at 1MHz with his "best" toroid.? (See "Experiments with Coils and Q measurements" amongst several articles he published).
PeterS??? ??? G8EZE
-----Original Message-----
From: Steven Greenfield AE7HD via groups.io <alienrelics@...>
To: [email protected]
Sent: Tue, 9 Aug 2022 13:02
Subject: Re: [Test Equipment Design & Construction] Making a Q-meter /
I don't think I've seen anyone mention the impact of the variable capacitor on Q. I understand that for lower Q, the losses in the inductor swamp losses in the capacitor. But at what point do capacitor losses begin to impact the measurements? In other words, will the garden variety tin coated, brass, or aluminum plate variable do? Does it need to be silver plated for lowest losses? Steve Greenfield AE7HD
_._,_._,_
|
The good old trick for fast sampling is to use a latching comparator. Those capture the comparator output at a time of an external latching pulse, and are usually very fast as well (usable well below 1ns).
Let the MCU timer + an external adjustable delay place the latching pulse at a desired time, and scan the threshold voltage repeatedly to find the voltage at that point in the sampled waveform. The waveforms are repetitive after all, so it’s a perfect application for such a sampler. It’s almost infinitely simpler to implement than a fast sample and hold, and works up to a couple GHz of you’re careful in the layout.
Part of the trick is to *measure* the timing of the latch pulse to close the loop around it. For that, the time-to-digital converter chips from eg. TI work great. They also make awesome reciprocal frequency measurement timers. And they all use an external reference frequency source to self-calibrate, so there’s little post-processing needed on the software side.?
I’ve put together several such samplers for various purposes (have to write them up eventually), and very lame MCUs can be used - the measurement just takes longer. In all of them, the job is done by a latching comparator, a voltage- or digitally-controlled delay element, a threshold voltage DAC, a time-to-digital converter, and an MCU timer peripheral.
The TTD converters these days are much less jittery than the low-end MCU GPIO pins and simple delay circuits, so you can measure the remnant closed-loop timing error of every sample and its contribution to the variance of the final result.
A suitable buffer + several comparators working in tandem can be used to speed up the sampling, but it gets expensive rather quickly, so an actual sampler is attractive if you’d need more than just one comparator per sampled channel.?
Cheers, Kuba
toggle quoted message
Show quoted text
8 aug. 2022 kl. 10:04 fm skrev Dr. David Kirkby, Kirkby Microwave Ltd <drkirkby@...>:
?On Mon, Aug 8, 2022 at 04:18 AM, tgerbic wrote:
I built a Q meter based on the ring down method that used an atmel processor. It was originally in the Silicon Chip magazine but you need a subscription to see the back issues. It was republished in EPE magazine in October and November 2007 and the PCB could be gotten from them back then. I etched my own. It works pretty well but is designed to read no higher than a Q of 120 mostly due to processor tech available. I suspect it could be expanded to higher values now that faster processors are available.? One nice thing is that the source code is available.
The EPE articles are on archive.org so you can see how the author designed it.
Thank you. I guess if one uses a fast sample and hold, the data acquisition could be done at a slower rate, as there's very little bandwidth of the signal. So there would be no need to sample at twice the resonate frequency of the coil. That's still quite a bit of work though - at least it would be for me, as not really an area I am good at.
|
On Tue, 9 Aug 2022 at 13:02, Steven Greenfield AE7HD via <alienrelics= [email protected]> wrote: I don't think I've seen anyone mention the impact of the variable capacitor on Q. I understand that for lower Q, the losses in the inductor swamp losses in the capacitor. But at what point do capacitor losses begin to impact the measurements?
In other words, will the garden variety tin coated, brass, or aluminum plate variable do? Does it need to be silver plated for lowest losses?
Steve Greenfield AE7HD
Somewhere I read that the capacitor loss can become significant above 10 MHz. So cheap capacitors would be yet another source of error. I would guess that a butterfly capacitor, where there’s no sliding contact would be preferable. Likewise silver plating.?
I can’t help feeling that the answer to the losses might be to introduce a further known loss, to determine the original unknown loss.?
Take for example an analog multimeter (eg AVO). If that is measuring a voltage from an unknown source impedance, and the impedance of the multimeter is unknown, you don’t know the source voltage, as there are two errors. I would have thought that using a couple of known resistors, one could deduce the internal resistance of the multimeter and the source resistance, allowing the source voltage to be determined.?
Dave? -- Dr. David Kirkby, Kirkby Microwave Ltd, drkirkby@...Telephone 01621-680100./ +44 1621 680100 Registered in England & Wales, company number 08914892. Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom
|
I don't think I've seen anyone mention the impact of the variable capacitor on Q. I understand that for lower Q, the losses in the inductor swamp losses in the capacitor. But at what point do capacitor losses begin to impact the measurements?
In other words, will the garden variety tin coated, brass, or aluminum plate variable do? Does it need to be silver plated for lowest losses?
Steve Greenfield AE7HD
|
On Tue, 9 Aug 2022 8:44? Dr. David Kirkby wrote:
"One
problem is the loading of the detector circuit. One can't avoid that,
although I would have thought R15 (4.7 M ohm) could be higher in value.
As far as I can see, it's only purpose is to stop the gate voltage
drifting off to an arbitrary value, with nothing to constrain it. Why
not make it 100 M ohm, as I have seen in one design."
- - - - - - - - - - - - - - - - - - -
The input capacitance of the detector can play a significant part at higher frequencies.? If the losses are low, it is not significant for Q measurements, but any variability with frequency/input level affects the measured inductance.
HP's method to control this was to use a high quality small series capacitor and a large-ish shunt capacitor as a capacitor divider to stabilise the strays around the high impedance end of the inductor and buffer the effects of the subsequent active components.
PeterS??? ??? G8EZE
|
They do appear from time to time… There are quite a lot still about. ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Dr. David Kirkby, Kirkby Microwave Ltd
Sent: 09 August 2022 08:10
To: [email protected]
Subject: Re: [Test Equipment Design & Construction] Making a Q-meter /? On Mon, 8 Aug 2022 at 12:38, nigel adams via <bristol.rell6l=[email protected]> wrote: Marconi Instruments also made a well respected Q Meter (TF1245) along with accompanying Oscillators (TF1246 & 1247), they were not far behind Boonton in this and published their own booklets on Q Meters and principles of measurement.
Worth a look, if anyone needs to see the manuals they are on the various archive sites as well as ours.
Just my 2p worth
Regards
Nigel Adams - Marconi Instruments Heritage Collection
I did place a bid one of the Marconi instruments, but the highest bidder bid ?1 more than me. There's another on eBay UK, but that lacks a source, and the source is unusual as the output impedance is low.
|
A few ideas were presented by Lloyd Butler VK5BR (Revised 2004), on how a simple analogue Q meter (LF & HF to approx. 40MHz), could be realised and descriptions of design problems and resolutions.?
An RF signal generator being used as the signal source with a suggested amplifier design to raise signal generator output signal levels to required value. Employs a wideband voltage follower output stage which develops a voltage across a 0.2 ohm resistor.
Measurable Q up to ~ 500.
See www.
John_M0JES
I think I will build something quite similar to that, just to see what's achievable in a simple design, although I would probably look for an HP Q-meter. My thoughts were.
1) Have an input for an external signal generator. I do have two, but the only one I have below 10 MHz has only push-buttons for setting the parameters, which I find a bit annoying. A rotary control would be a lot nicer.
2) Use an internal oscillator. I can't actually find any that suit my desires very well. The one in PW is not synthesised, but that might do at least as a starting point.
3) Add the amplifier given in the circuit above.
4) Add the rest of the circuit above.
That would give a fully self-contained unit, with an option to use an external source.
The design above says 5 cm of wire presenta? a significant inductance, so impedance at higher frequency. But it seems to me that the particular part of the circuit should be mounted very close to the connectors. Then we would have much shorter leads, and that could be a wide strap, reducing the inductance.
One problem is the loading of the detector circuit. One can't avoid that, although I would have thought R15 (4.7 M ohm) could be higher in value. As far as I can see, it's only purpose is to stop the gate voltage drifting off to an arbitrary value, with nothing to constrain it. Why not make it 100 M ohm, as I have seen in one design.
Since one can't avoid the fact that the detector adds some load, it got me wondering if it's possible to apply a few fixed loads (maybe 100 k ohm, 1 M ohm, 10 M ohm, and use the change in voltage at the output of the detector to estimate the amount of loading the detector is actually causing to the tuned circuit. Or maybe purposely make the detector have a lowish (say 100 k ohm) impedance, then correct for the effect it is having. It would mean one could not drive a meter directly from the detector to make a Q-measurement, but either have only a digital readout (not my preference), or drive an analog meter with a voltage from out of an DAC. So have
* ADC make a measurement of the output level of the detector
* CPU doing some processing to correct for all known error sources * DAC driving an analog meter.
Dave
|
Dr. David Kirkby, Kirkby Microwave Ltd
Jempi ( ON7MA )
John Kolb
Pete_G4GJL