|
Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members. |
|
Thread Tools |
27th Jan 2018, 11:18 am | #21 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Compensating for VFO drift
That’s an interesting idea as well. This is fun
|
27th Jan 2018, 12:02 pm | #22 | |
Rest in Peace
Join Date: Sep 2008
Location: Solihull, West Midlands, UK.
Posts: 4,872
|
Re: Compensating for VFO drift
Quote:
|
|
27th Jan 2018, 12:20 pm | #23 | |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Compensating for VFO drift
Quote:
I think you are mostly looking at alteration in inductance with increasing temperature. Looking at the picture of the coil, it is not constructed on a rigid enough former, it doesn't look like it, but its hard to tell. The geometry of the coil needs to be super rigid and preferably on a glass or ceramic form and wound very tightly. It only takes microscopic changes in the geometry with temperature to cause significant drift. It it less likely your NPO caps are the problem and you have a stable power supply. You could try heating the transistor, if that gave an effect in the drift direction, you could increase its DC emitter degeneration can help a bit. You might be better off with a physically larger coil wound with thick spaced tinned copper wire that is pre stretched to stiffen it too. Surplus Sales Nebraska has quite a few ceramic coil forms. If you throw in a core of some kind, it is another variable. And as noted, the osc amplitude and feedback drive are issues. If too low unreliable starting, if too high, drifts are amplified too and more harmonics in the output. |
|
27th Jan 2018, 1:39 pm | #24 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Compensating for VFO drift
The coil that was wound was a sheet of paper doused in superglue. They are similar in strength to phenolic in theory when set. I used to make fireworks as a kid using the same method to make a louder bang .... Not sure what it does to the Q but it is rock solid. Bashing the VFO around doesn’t noticeably change the frequency.
I've just picked up a dead Tek scope so I’ll be distracted by that for a few days no doubt but will post back with more findings when I do some more experimentation. |
27th Jan 2018, 4:25 pm | #25 |
Heptode
Join Date: Nov 2009
Location: London, Ontario, Canada
Posts: 675
|
Re: Compensating for VFO drift
Wes Hayward had an excellent article in the '93 QST magazine, pg. 37, titled "Measuring and Compensating Oscillator Frequency Drift".
Doing a quick search, it can be found on the web too: http://www.robkalmeijer.nl/techniek/...e37/index.html Peter |
27th Jan 2018, 4:43 pm | #26 |
Dekatron
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 14,007
|
Re: Compensating for VFO drift
When I had to build VFOs I found the recipe for success was to use lots of capacitance [so swamping the effective capacitances of the active device] and a coil with very few turns (so lowering the resistive losses and increasing the 'Q')
Grooved ceramic formers are a good thing: the likes of the WWII Collins TCS12 receiver used them - with the wire wound into the grooves under tension to provide maximum mechanical stability. Another useful trick when using JFETs was to dispense with the usual gate-resistor/capacitor bias and instead fit a (bypassed) source-resistor to provide the bias. That way the gate is always driven at a low impedance (again reducing the effect of changes in the active device's internal impedances). Then the final tweak: make the source-bias resistor variable, and adjust it so that the source-current is the same whether the thing is oscillating or not. This "Class-A" oscillator is sometimes called a "Q-gate" oscillator and works remarkably well in both FET and vacuum-tube versions. |
27th Jan 2018, 7:41 pm | #27 | |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Compensating for VFO drift
Quote:
On the other hand, unlike the VFO oscillator tank circuit, a heavily loaded tank circuit in a power output stage the Q is R/X, where R is the load resistance, so in that case the Q could be made to increase by lowering the reactances, but this is not normally the case in a VFO where the loading on the resonant circuit should be low. |
|
27th Jan 2018, 7:58 pm | #28 |
Dekatron
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 14,007
|
Re: Compensating for VFO drift
I guess it depends on the extent to which in the real-world the capacitors and inductors have an inescapable resistive-component.
And the extent to which putting lots-of-C across the intrinsic-capacitance of the active device in the oscillator helps isolate the frequency-determining part from the impedances and loads of the gain-part. Vackar, Gouriet, Seiler etc all put lots-of-capacitance across the active-device's intrinsic capacitance, then use a series-tuned circuit with small values of L and C to do the frequency-determination |
27th Jan 2018, 8:10 pm | #29 | |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Compensating for VFO drift
Quote:
|
|
27th Jan 2018, 8:22 pm | #30 |
Hexode
Join Date: Jul 2014
Location: Magor, Monmouthshire, Wales, UK.
Posts: 436
|
Re: Compensating for VFO drift
I was wondering who was going to mention the Vackar config.
From my own experience I can recommend it, given that the other variables such as construction style are the same, I've had way better results from vackar oscillators than anything else. I've found the Hartley to be the most reliable with low gain and marginal BJT & FETs. Here is some more reading http://www.f6evt.fr/f6evt_fr/vackar_wholepaper.pdf T
__________________
Adapt, Improvise, Oh Bother..... Last edited by Tyso_Bl; 27th Jan 2018 at 8:22 pm. Reason: layout |
27th Jan 2018, 11:03 pm | #31 | |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Compensating for VFO drift
Quote:
There is an extensive oscillator analysis in Terman's Rado Engineers's handbook. Terman also shows 15 different oscillator circuit configurations( Colpitts, Hartley feedback, reverse feedback, unity coupled etc) where, if the reactive values are split and proportioned correctly for each case, the oscillation frequency becomes independent of the valve constants. This would obviously work with a Fet too. Ideally for a VFO, its center frequency would be set up so that these conditions were met, then apart from any other considerations a significant source of frequency drift is eliminated. I scanned this page once and posted it before on another thread. |
|
29th Jan 2018, 6:57 pm | #32 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Compensating for VFO drift
I'm going to build a Vackar next time. Did some modifications to this one today and it's nice and stable now:
1. Changed coupling caps to 10pF 2. Added varactor pair for tuning. 3. Added 10 turn pot for voltage source for tuning. 4. Added bandspread trimmer This hits 6998-7042 on the mark! Total drift is around 8Hz/minute averaged over 2 hours. Now to add a buffer amp, filter and stick in a box so it isn't quite as dicky with breezes etc. |
29th Jan 2018, 11:38 pm | #33 | |
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Compensating for VFO drift
Quote:
I enjoy hand-winding coils, so I may get drawn into a similar project once the current major one is off the drawing board (... ie, nowhere close to soon!) I was wondering about stability in the good old golden days of radio, pre and post-War. I wonder what they did to get this right in the big old transmitters of that era? My pre-War Marconi (receiver ) is pretty stable, even now, but I haven't measured drift in any analytical way.
__________________
Al |
|
30th Jan 2018, 1:06 am | #34 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Compensating for VFO drift
Yes they certainly are rather fun to wind. Could be addictive. I think it's because you've actually made your own electronic component from scratch. One thing I find really fun is measuring them afterwards with a sig gen and scope. Looking at it, the coils were larger and wound with thicker wire. I assume that leads to thermal and mechanical stability.
I've included a picture of the horrible lash up here: If anything I'm actually slightly surprised at home much better an air core coil that probably cost me a few pence at most outperforms commercial toroids by an order of magnitude. As always though, the hopeful cost cutting escape from the air variable capacitor was a failure as a decent 10 turn Bourns pot costs as much as one of them did If anyone has a TF930 counter and finds this via google, the capture software is here (C#). Writes out a CSV you can stick in Excel containing two columns: seconds from program started, frequency (hz) Code:
using log4net; using log4net.Config; using System; using System.Diagnostics; using System.IO; using System.IO.Ports; using System.Text.RegularExpressions; using System.Threading; namespace DriftCapture { class TF930 : IDisposable { static ILog Log = LogManager.GetLogger(typeof(TF930)); SerialPort port; Regex parseValueRegex = new Regex("^([0-9.e+]+)", RegexOptions.Compiled); public TF930() { Log.Info("connecting to TF930"); port = new SerialPort("COM3", 115200, Parity.None, 8); port.Open(); port.WriteLine("*IDN?\n"); var identity = port.ReadLine(); Log.InfoFormat("Identified TF930: {0}", identity); Log.InfoFormat("Resetting counter"); port.WriteLine("*RST\n"); } public void Dispose() { if (port.IsOpen) port.Close(); port.Dispose(); } public double ParseResult(string data) { var y = parseValueRegex.Match(data); return double.Parse(y.Value); } public double MeasureHz() { Log.InfoFormat("Measurement in KHz requested"); port.WriteLine("N?\n"); var z = ParseResult(port.ReadLine()); Log.InfoFormat("Got: {0}Hz", z); return z; } } class DriftWriter { Stopwatch stopwatch; FileStream fs; StreamWriter sw; public DriftWriter(string path) { stopwatch = new Stopwatch(); fs = File.Create(path); sw = new StreamWriter(fs); stopwatch.Start(); } public void WriteEvent(double value) { var toWrite = string.Format("{0},{1}", stopwatch.Elapsed.TotalSeconds, value); Console.WriteLine(toWrite); sw.WriteLine(toWrite); sw.Flush(); } } class Program { static void Main(string[] args) { var dw = new DriftWriter("drift.csv"); BasicConfigurator.Configure(); var counter = new TF930(); for (; ; ) { dw.WriteEvent(counter.MeasureHz()); Thread.Sleep(10 * 1000); // 10s } } } } |
30th Jan 2018, 1:36 am | #35 | ||
Dekatron
Join Date: Nov 2006
Location: London, UK.
Posts: 3,496
|
Re: Compensating for VFO drift
Quote:
I wouldn't knock it too much. Sometimes it's just practical and efficient to quickly prove or disprove a theory. I don't normally do dead-bug but I do use copper sheet and 'Manhattan' style stick-on solderable pads which are from the QRP community. I really like your picture as it gives an insight into the process that got you from question to coding! Quote:
In a completely different context, with heavy DC and RF currents, it seems to be that RF chokes between the circuit and power supply are better when air-wound; less saturable...
__________________
Al Last edited by Al (astral highway); 30th Jan 2018 at 1:46 am. |
||
30th Jan 2018, 2:01 am | #36 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Compensating for VFO drift
I tried Manhattan style for a few things but I kept getting the little pads stuck to myself like a 4 year old
4.4uH calculation to 4.7uH measured here. That’s 6.4% it which is pretty good. The band spread cap tunes that out nicely. I had a home constructed amplifier given to me once that had massive RF chokes in it that were air wound. Definitely going to make more inductors now. After reading your other posts perhaps a small Tesla coil is in order for the living room table here |
23rd Feb 2018, 7:00 pm | #37 |
Octode
Join Date: Jun 2013
Location: Huddersfield, West Yorkshire, UK.
Posts: 1,795
|
Re: Compensating for VFO drift
Hi Peeps.... Just an addition to the "Manhattan" style of construction..I needed a dual opamp in the mix... so I knocked these little modules up. May be of use to you.
__________________
Should get out more. Regards Wendy G8BZY |
24th Feb 2018, 1:15 am | #38 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Compensating for VFO drift
That's rather neat looking. I tend to be lazy and stick them upside down and solder the ground pins to the ground plane.
Meant to update this post actually. I have started to procure bits to build an environmental chamber. After doing some research it appears the best approach for this is bonding a peltier heat pump to small metal biscuit tin and a CPU heatsink/fan for cooling. For heating, stuffing a few 10 ohm 3W resistors in will do. This is then inserted inside a larger metal biscuit tin padded out with polystyrene with the cooler sticking out of the top. An umbilical would provide some clip leads which can be stuffed on the feed through capacitor leads. A 40mm fan would be inside the smaller tin to circulate the air inside it. Procurement has unfortunately only got as far as buying two tins of biscuits and starting consumption of them and scribbling out ideas for control systems. This lead to the discovery that I was out of tea bags Control of the temperature is as yet undecided but schemes of an arduino driving logic level MOSFETs driving the peltier and resistors is looking most sensible. Measurement of temperature can be done with an LM335Z I found lying around. |
24th Feb 2018, 7:24 am | #39 |
Heptode
Join Date: Oct 2005
Location: Olympia, Washington, USA.
Posts: 664
|
Re: Compensating for VFO drift
How about one of the 4 pin sealed xtal units like they use in computers? They seem pretty stable and you could use a divider IC for changing the freq, output.
IIRC, Digikey or Mouser sells them. |
24th Feb 2018, 9:32 am | #40 |
Dekatron
Join Date: Apr 2012
Location: Wiltshire, UK.
Posts: 14,007
|
Re: Compensating for VFO drift
Thinking about heating your chamber (ooer Matron!) I'm reminded of a design for a crystal-oven I saw ages ago where rather than having resistors to provide the heat, it actually used a couple of medium-power transistors (BFY51 sort of thing) as the heating 'element', wiggling the base bias to vary the heating effect. There was a simple 741 as control (with variable loop-gain to adjust the speed of response) and a couple of forward-biased diodes provided thermal sensing of the crystal temperature.
|