|
Vintage Test Gear and Workshop Equipment For discussions about vintage test gear and workshop equipment such as coil winders. |
|
Thread Tools |
13th Dec 2016, 11:54 pm | #21 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: R.F. voltmeter: your thoughts, please
I'm sorry that I don't feel competent to respond to those specific questions, but... this is my latest design find in this area which may be of interest http://www.mw0llo.com/milliwatt.aspx
B Last edited by Bazz4CQJ; 14th Dec 2016 at 12:05 am. |
16th Dec 2016, 1:07 am | #22 |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: R.F. voltmeter: your thoughts, please
Ahem! Now that I've found the time to read in depth the excellent write-up by the author, I can answer my own question! The author does indeed refer to a +ve supply rail as an alternative and the components which require polarity reversals are indeed the ones I have listed.
It is my intention to build this unit (+ve supply rail version); it's just a question of when. Aside. As for the write-up itself, it is quite clear that the author, (at the time of writing) was professionally engaged in electronics. I also note that his address is given as Palo Alto, California. Al. |
16th Dec 2016, 4:03 pm | #23 |
Nonode
Join Date: Jan 2009
Location: Bristol, UK.
Posts: 2,059
|
Re: R.F. voltmeter: your thoughts, please
I think replacing the tail resistor with a current source is only necessary if there is a lot of common mode voltage (to the FETs) with respect to the supply voltage. Possibly the probe cable may pick up a lot of common mode induced 50Hz. For the sake of another FET and a resistor it must be worth trying!
dc |
17th Dec 2016, 2:32 am | #24 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: R.F. voltmeter: your thoughts, please
I thought I would weigh in on the technical side of things about the voltmeter. I agree with the other posts. The main problem with it is that at low signal levels the input rectifiers have significant non-linearity, hence on the low ranges the non linear and compressed meter scales, not very nice. So the problem is really non linear low level RF signal detection, and how to fix that:
There have been some interesting ways to fix this in the past, one trick is to apply a small amount of bias to the diodes to help overcome the diode voltage drops a little, this doesn't give a perfect result. Another tick is to use a diode in a feedback loop around an OP amp to make a precision detector. All electronic detectors that run active devices from some power rails will have a limited dynamic range and fail with very large signals at some point, unlike a single diode. Yet another trick for small signals: use a Tunnel diode in reverse, called a "back diode" it acts as though it has no forward voltage drop. One of the cleverest circuits I have seen was presented in the Art of Electronic by Horowitz & Hill. This is a very clever idea. Instead of driving a detector diode with a voltage source, you drive it with a current source.Since a current source attempts to maintain a current, the voltages rises as high as it needs too, to establish forward current through the detector diode, and this overcomes the diode's non-linearity. Horowitz & hill used a transistorized current source to drive the detector diode. With the above in mind it occurred to me that instead of using a transistorized circuit to make a current source, why not just use an inductor. Inductors act a lot like current sources in that they attempt to oppose a change in current. So I cooked up a circuit that became known as SUPADETECTOR, it got published in Silicon Chip. I have attached the graph and circuit which compared this detector to a 1N60 diode. On the face of it, it looks like a normal detector with an impedance mismatched drive, until one understands the principle of it and the uniformity of the output due to the damping. The transistor converts the incoming signal into a current and the inductance attempts to oppose changes in the value, so the voltage across the detector diodes climbs as high as required to overcome the diode's non-linearity. I put these in as detectors into my Eddystone EC10 radios and other radios like the OC16/AF178 radio and others I have made and it profoundly improves the detection of low level signals. The principle can also be used for metering. Here is a link to the AF178/OC16 radio for those who like germanium home brew radios: http://worldphaco.com/uploads/THE_AF...OC16_RADIO.pdf I also saw on another thread suggestions for replacement for AF114-7 transistors. Not mentioned there, the AF178 (although intended for TV work) is by far the better choice I think after performing many gain & noise tests, and AF185 is also a reasonable option. Also to replace the AF118 use the 2SA358, there are few options. Last edited by Argus25; 17th Dec 2016 at 2:52 am. Reason: read better. |
17th Dec 2016, 4:23 am | #25 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: R.F. voltmeter: your thoughts, please
Clearly, the Supadetector works well when used on a 455kHz transistorised IF stage, but I guess it might be less useful if used on a valve TV IF stage.
In considering RF probes, there are probably optimum options for specific tasks, but the germanium point-contact diode does seem to win the prize for versatility; if you could use only one probe, then that might be the one. Alternatively, Schottky diodes like the 1N5711 seem to offer similar performance (especially if biased) and may be easier to obtain and match. B |
17th Dec 2016, 6:48 am | #26 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: R.F. voltmeter: your thoughts, please
Bazz4CQJ,
Yes, you are right, the signal levels on a valve TV IF output are not much use if they are so low they are down in the snow and recovery of a low level signal isn't much help for the sync and vision circuits. But in transistor radios, listening to low level demodulated audio, the supadetector does help with much lower distortion than a plain diode. |
17th Dec 2016, 11:51 am | #27 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: R.F. voltmeter: your thoughts, please
..more on metering low level RF signals..
When I built my "Alexandra Palace Transmitter in a Box" apart from having crystal controlled modulators, I figured it would be good to have accurate metering (and level control) of the peak RF output level. I was not happy that the level was not known for other converters/modulators. Plus it would be useful when setting up the tuning in restored 405 line sets, or if I built any 405 line sets like the Argus. I have attached a circuit that shows how an accurate level meter can be made for low level signals. They are first amplified by monolithic RF amp IC's and then sent on to an Avnet UTD1000. (these are a moderately expensive item, but they work well) then to a meter using an OP amp driver. Its worth looking up the spec sheet for the UTD1000. |
17th Dec 2016, 12:06 pm | #28 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: R.F. voltmeter: your thoughts, please
I found the data sheet, the UTD 1000 is good from 10 to 1000 MHz.
http://www.teledyne-cougar.com/pdfs/...td10001001.pdf |
17th Dec 2016, 1:25 pm | #29 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: R.F. voltmeter: your thoughts, please
That's a device that's new to me; the comparable AD8307, discussed earlier in the thread, seems to have attracted a lot of attention for amateur use, perhaps arising from its use in a circuit published in QST some time ago, which has been followed by a number of derivative circuits since then. It would be interesting to see a side-by-side comparison of the specs of the two chips. As per Terry's post, working AD8307's are being offered very cheaply out of China .
B |
18th Dec 2016, 8:24 am | #30 |
Nonode
Join Date: Oct 2010
Location: Tintinara, South Australia, Australia
Posts: 2,324
|
Re: R.F. voltmeter: your thoughts, please
The AD8307 is spec'd to 500MHz, with some usage up to about 650MHz with reduced accuracy.
For those with good eyesight and steady hands, there is the AD8317, good from 1MHz to 10GHz. Terry |
18th Dec 2016, 2:16 pm | #31 |
Moderator
Join Date: Jan 2003
Location: Ipswich, Suffolk, IP4, UK.
Posts: 21,192
|
Re: R.F. voltmeter: your thoughts, please
Posts relating to fake components moved to this thread:-
https://www.vintage-radio.net/forum/...d.php?t=122038
__________________
Graham. Forum Moderator Reach for your meter before you reach for your soldering iron. |
18th Dec 2016, 5:39 pm | #32 | |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: R.F. voltmeter: your thoughts, please
Quote:
B |
|
18th Dec 2016, 5:42 pm | #33 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: R.F. voltmeter: your thoughts, please
Quote:
I've only looked at the detector in a basic sense but it appears to use two jellybean silicon diodes with a bias current of a few uA. For very small signals in the mV range this biased diode arrangement will look like a resistor. There will also be a fair bit of diode self capacitance when it is biased like this. You can get a guesstimate of the resistance at lowish frequencies from a simple equation where R = n*Vt/I where n is typically 2 for a silicon diode and Vt is the thermal voltage ( = energy per carrier charge in the doped silicon at room temperature = kT/q = 0.026V) So if the bias is 2uA then the diode resistance will be 2 * 0.026/2e-6 = 26k ohms.There's two diodes in the detector that will appear as if they are in parallel so the source will see 13k ohm Rp for small signals at low frequency. As you move up into the VHF region this Rp will fall considerably and you could be looking at an Rp of maybe 1-2k ohm here. This is going to be fine for probing low Z sources, eg a 50 ohm signal path but I think the probe could load high Z circuits if the signal level is very small. Up at higher signal levels I think the loading will be far less because the above equation only applies to tiny AC signals. So the input impedance will appear to be level dependent. Very small signals down at a few mV will get loaded the most.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 18th Dec 2016 at 5:55 pm. |
|
18th Dec 2016, 5:48 pm | #34 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: R.F. voltmeter: your thoughts, please
The logarithmic detectors with wider bandwidths suffer from reduced sensitivity. The wider bandwidth equates to a wider band of total noise power, so their noise floor rises appreciably.
It's best not to use any of the 3, 6, 10GHz ones unless you really need them. The 500MHz family are a lot better performers below 500MHz. The old way of putting it was "The wider you open the window, the more **it flies in!" Oh, and if you go for the 500MHz part with the limited outputs, you can use one of them for a counter output, then your counter will go to signal levels as low as your voltmeter meter does. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
18th Dec 2016, 10:16 pm | #35 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: R.F. voltmeter: your thoughts, please
I have attached a couple of circuits as they make some interesting points about RF metering.
One is an RF meter on part of an AM transmitter system that goes up to a carrier frequency of 7.5MHz. Since the system impedance was fixed at 50R, and since power is proportional to the square of the voltage, if the voltage is squared, the output of the squarer is proportional to power. The squaring was done with an AD734 as shown in the red box on the right of the diagram. An AD633 is a cheap but good IC that can also be used as a squarer, mutiplier, modulator etc in the lower frequency ranges. On the left, the red box shows a way of getting rid of a rectifier's voltage drop, by getting one transistor junction to cancel the other with an emitter follower pair, this can replace the BAT49 diode. The other circuit is for a 0-30w compact VHF directional watt meter for testing VHF handhelds feeding their antennas, which shows forward and reverse power. The directional coupler is simply made from a short length of coax. It uses a lot of the techniques mentioned. The input diodes are forward biased, the detector is linearized by an OP amp with diode feedback, the rectified signal is squared by an NJM4200. The advantage of this is that once the signal becomes DC at the output of the coupler, the frequency response of the other meter parts is not important and a coupler can have a very wide bandwidth. (On a historical note; the 4 quadrant analog multiplier was patented by Gilbert in the '70's and it ended up getting used on the inputs of Tek scopes. Though the configuration was invented by Jones in the '60's, but he never got the credit due to the obscurity of his work. The circuits can be used for squarers, dividers, mixers,modulators, phase detectors and many more. The common IC most people know with the crossed collector configuration is the MC1496, often used in color TV work) |
18th Dec 2016, 10:48 pm | #36 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: R.F. voltmeter: your thoughts, please
In ALC circuits, the gain-controlled device is often run in a mode where the logarithm of power gain (dB) is a straight line when plotted against control voltage over the region used. It's useful to pair this with one of the logarithmic RF detector devices because then the loop gain of the ALC syste stays constant over level, so the time constants don't seem to wander and stability is a bit easier to assure.
Back in about 1990, G3ROO and myself built a couple of directional power meters based on the old AD6440 logger, one for forwards, one for reflected. Mine has two meters reading forward and reflected power on linear scales of dBW. A meter floating, measuring the difference between the logger outputs is equivalent to having a divider in non-logged signals, so the meter reads a linear scale of dB of return loss, which can be marked with a nice expanded-law VSWR scale. On the principle that if a job's worth doing, it's worth overdoing.... I added a peak hold for SSB and a phase detector looking at the limiter outputs. So my meter works as a vector network analyser. I get S11 in magnitude and phase so I know what antenna impedances really are. The AD633 is a true Gilbert cell, with the linearising diodes in place. Barrie worked for Plessey, then Tek, then ADI. ADI are said to have started a new division near his home just to get him on their team. I think the investment has paid off. James Bryant G4CLF of the 9MHz plessey chip IF module fame is another ex-Plessey guy who wound up as ADI's European marketing boss. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
19th Dec 2016, 5:34 am | #37 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: R.F. voltmeter: your thoughts, please
David,
About 2 years before Gilbert patented the "cell", Nagra in Switzerland produced a tape deck (it might have been the MK2) these were used extensively by the BBC. Anyhow, if you look at the phase detector in there, for the capstan servo system (the capstan could lock to a crystal reference to stop lip sync drifting off), it is the Gilbert cell. So both Jones and Nagra used it beforehand, but Gilbert patented it, it seems. One super convenient use of the 4 quadrant multiplier, as used in Tek scopes, is the ability to invert a signal, with a DC control, like invert channel 2, discussed on page 4- 5 of this article: http://worldphaco.com/uploads/TEKTRO...ILLOSCOPES.pdf Last edited by Argus25; 19th Dec 2016 at 5:43 am. Reason: read better |
19th Dec 2016, 3:37 pm | #38 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: R.F. voltmeter: your thoughts, please
Surely, this thread is getting rather off-topic?
B |
19th Dec 2016, 4:43 pm | #39 |
Dekatron
Join Date: Apr 2009
Location: Oxfordshire, UK.
Posts: 4,923
|
Re: R.F. voltmeter: your thoughts, please
I don't have the insight to comment on the pros and cons of the amp circuit in the OP, irrespective of whether it's done with P or N channel fets. However, can't this function be achieved very well using a CA3140 chip (or something similar) which has a high-impedance front end (MOSFET?) and appears in various circuits which can be found doing exactly this job of RF voltage measurement (using a diode probe front end)?
Can anyone suggest what the downsides to that approach are? B |
19th Dec 2016, 6:38 pm | #40 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
|
Re: R.F. voltmeter: your thoughts, please
There are plenty of high-z opamps, and they would offer more stable balance and gain than the FET and bipolar pairs design is capable of. You'd probably still want an offset = zero adjuster to trim diode imbalance. Being me, I'd stick in a different gain trimmer for each range, then I wouldn't have to worry about finding a compromise for the combined ranges.
A bit of antistatic protection with some clamp diodes might save it from destruction every now and then. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |