Test Equipment Anonymous

These are generic RF signal generators I've seen around since the 90s. This one is branded Maplin, but they were also branded Rapid, Tenma, Altai, Leader and others. There have also been very similar published amateur designs. I'd always wondered about them and noticed they were surprisingly expensive. People seem to like small, neat, modern boxes. I saw it for a tenner at a swapmeet and bought it, mainly with the intention of flipping it.

It's a Colpitts oscillator with six ranges covering 100kHz to 150MHz. The highest range is also labeled as Harmonic 96-450, so it's claimed to be usable on that range using the third harmonic.

I thought it might be interesting to play with and I ought to make sure it worked before selling it.

The frequency accuracy is given as +/- 3% and the output 100mV (approx) up to 35MHz. It doesn't quote an output impedance, but I'd estimate it varies over a range of a few hundred Ohms. I haven't tried measuring it. It isn't a standard 50 Ohm output. It doesn't have a proper attenuator, it has a pot and a switched potential divider giving High and Low ranges.

It has internal AM modulation at about 1kHz and can be externally modulated. There's also a crystal position and sockets for a of crystal up to 15MHz, on the front. The manual discusses how this can be used with an external receiver for accurate "calibration". The inductors and capacitors have no trimmers and the unit is stuck with the calibration it had when it left the factory.

I tried connecting it to a scope and it was more or less accurate. It does have a slow motion drive but the cursor is quite coarse. There are markers on the scale for 455kHz, 4.5MHz and 10.7MHz, all common IF frequencies. The output amplitude is far from constant and is very much lower on the higher ranges. It rises and drops noticeably across the upper scales and harmonics are prominent. I'd guess this is because of stray capacitance and inductance. Probably the most irritating aspect of using it for IF alignment would be that it's hard to set and reset with sufficient accuracy, even using the markers. This can't be done conveniently from the output using a frequency counter. There is a modification which takes the signal from where it enters the output attenuator, through a capacitor, to a BNC socket. So a frequency counter will have an adequate signal. Many frequency counters have an oscilloscope type 1MOhm 20pF input on their low range. The modification is straightforward and so I did it.

With the modification in place it's much easier to set the frequency accurately. For instance it can be set to 455kHz to within about 50Hz without much trouble. It drifts surprisingly little.

Circuit diagram. Sig Gen connected to scope and frequency counter.

Internals of sig gen. Scope screen shot at 455kHz. Scope screen shot at 20MHz.

This is definitely a hobbyist instrument. It lacks internal screening and leakage can easily be picked up on a scanner. The output amplitude varies wildly across the upper ranges. The distortion of the sine wave on the higher ranges is striking. AM modulation is crude. I have an Advance RF sig gen from around 1960 which uses a single ECC81 (12AT7). It is not a lab grade instrument but it has much better screening than the Maplin unit and does have presets to adjust it. There are good reasons why lab standard RF sig gens are expensive.

I believe the Maplin XG94C works as well as it ever did and the modification makes it more useful. It could be used to do IF alignments on radios, which I'd guess was one of its major intended purposes. I must admit I'm not very impressed by it.

Interesting stuff. Apropos your distortion comment, take a look at an early synthesised generator (I'm assuming later ones are better). Ten years ago, I was horrified when I connected £2k-worth of synthesised RF generator to a wide bandwidth oscilloscope. As soon as I got home, I tried the same test on my E4420B into a 1GHz oscilloscope and was delighted to discover that its digits only extend as far as the control system, with the RF department being analogue. I'd certainly agree that the internal construction of that generator is Mickey Mouse. Built to a price and amateur expectations.

Zenith wrote: ↑Tue Apr 28, 2026 7:32 pm {quote]
The frequency accuracy is given as +/- 3% and the output 100mV (approx) up to 35MHz.

Reminds me of the "accuracy" of old shortwave/AM/FM receivers with the station name printed on the front panel tuning dial. I played with some as a bored schoolkid in the 70s, and was frustrated by the vagueness of the calibration, limited transmission times, and (of course) SW propagation.

That was the one time I wished I had a frequency counter, to calibrate the receiver.

In the mid 80s I bought one of the first consumer synthesised receivers, the Sony ICL 7600. Arthur C Clarke referred to his as a "spy radio", since it was small, accurate and had good performance.

It doesn't quote an output impedance, but I'd estimate it varies over a range of a few hundred Ohms. I haven't tried measuring it. It isn't a standard 50 Ohm output. It doesn't have a proper attenuator, it has a pot and a switched potential divider giving High and Low ranges.

Oh yuck. Maybe that accounts for your 20MHz screenshot, which looks like a square-ish wave with reflections. Do the minor peaks stay at the same time delay as you vary the frequency?

The output amplitude is far from constant and is very much lower on the higher ranges. It rises and drops noticeably across the upper scales and harmonics are prominent. I'd guess this is because of stray capacitance and inductance.

Or possibly VSWR impedance mismatch.

By the time I needed any generator, I was using decent professional equipment. Hence the concept of specifying a "levelled generator" always felt rather unnecessary to me.

I have since acquired one at the Harwell rally (RIP), a Tek 191 "constant amplitude signal generator". Ought to sell it (with a BNC-GR874 adaptor as an extra cost accessory )

This is definitely a hobbyist instrument.
...
There are good reasons why lab standard RF sig gens test equipment are expensive.

Maplin, and FTFY. Just feel the weight of RF TE!

I believe the Maplin XG94C works as well as it ever did and the modification makes it more useful. It c
ould be used to do IF alignments on radios, which I'd guess was one of its major intended purposes. I must admit I'm not very impressed by it.

For me its sole attraction would be to use my imagination to make a piece of awfulness do something useful.

If, long ago, I was a radio ham I might have been interested.

It's a nasty little thing, and the standard of construction and the thought that went into it, is well below Advance, Heathkit and other low end sig gens of its time or earlier. By the 50s HP, Marconi and others were making sophisticated RF sig gens, but they were huge and must have cost a fortune. A Marconi TF144 I had included a disk seal triode in a carefully engineered enclosure, it used a special RF double tetrode for output and had a stabilised supply using a KT66. The inductances were on an ebonite turret, which was itself a work of art. The HP8640A I used for years, is a magnificent instrument. It still works but the dacron gears have cracked, which is a common problem with them. When it was new it sold at a price which would easily have bought a house. The RF shielding in both of them is fanatical.

However, the Maplin sig gen sold for years under various badges, and people seemed happy with it for the price. I believe they are popular with vintage domestic radio enthusiasts.

tggzzz wrote: ↑Wed Apr 29, 2026 8:07 am

It doesn't quote an output impedance, but I'd estimate it varies over a range of a few hundred Ohms. I haven't tried measuring it. It isn't a standard 50 Ohm output. It doesn't have a proper attenuator, it has a pot and a switched potential divider giving High and Low ranges.

Oh yuck. Maybe that accounts for your 20MHz screenshot, which looks like a square-ish wave with reflections. Do the minor peaks stay at the same time delay as you vary the frequency?

I didn't notice any change in the time delay, I assumed the change in the waveform was down to the amplitude of harmonics changing as the fundamental was changed. I didn't really look and I have no intention of messing with it further. I was surprised by the change in output amplitude across the upper ranges. In one place it dropped to very low levels.

tggzzz wrote: ↑Wed Apr 29, 2026 8:07 am

I believe the Maplin XG94C works as well as it ever did and the modification makes it more useful. It c
ould be used to do IF alignments on radios, which I'd guess was one of its major intended purposes. I must admit I'm not very impressed by it.

For me its sole attraction would be to use my imagination to make a piece of awfulness do something useful.

If, long ago, I was a radio ham I might have been interested.

The main attraction was flipping it. I was curious as to how it would perform.

You'd have been better off with a used Advance H1 or similar. They are bigger, heavier and look old-fashioned, but perform much better. They were probably cheaper than these when they were new.

Zenith wrote: ↑Wed Apr 29, 2026 8:23 am It still works but the dacron gears have cracked, which is a common problem with them. When it was new it sold at a price which would easily have bought a house. The RF shielding in both of them is fanatical.

However, the Maplin sig gen sold for years under various badges, and people seemed happy with it for the price. I believe they are popular with vintage domestic radio enthusiasts.

You might be able to find/design the gears 3d printed in SLA resin. That has excellent smoothness and resolution, and is cheap. Ditto brass, except expensive All you need is the 3d model. If you have dimensions, there's a good chance there's an OpenSCAD parameterised library that will do most of the job for you.

And yes, the price comparison with a house is useful for getting people's attention!

tggzzz wrote: ↑Wed Apr 29, 2026 11:44 am You might be able to find/design the gears 3d printed in SLA resin. That has excellent smoothness and resolution, and is cheap. Ditto brass, except expensive All you need is the 3d model. If you have dimensions, there's a good chance there's an OpenSCAD parameterised library that will do most of the job for you.

And yes, the price comparison with a house is useful for getting people's attention!

Thanks for that. I'll look into it. I was aware someone had brass gears made a few years back, and for a time sold them on ebay. They were $200 as I recall. When the HP 8640A began to play up because of the disintegrating gears, I bought a Marconi 2019 from Stewart of Reading and I like the push button control of everything, so there wasn't that much of an incentive to fix the HP8640A.

I think quality RF TE has always been relatively expensive. The lowest spec Siglent and Rigol models, which go to 2GHz and 1.5 GHz, are about £2,000 inc VAT. The top end models which go up to 40GHz, are well over £30,000. I haven't looked at Keysight, R&S etc prices, but I suspect they are a lot more.

Yeah, life is too short to be bored.

I made this P6013 HV probe handle for ~£20 in SLA.


The thread is from an OpenSCAD library, with a 2mm pitch. I like OpenSCAD (and equivalents). The concept of adding/subtracting primary shapes matches the way I think about constructing things.

Zenith wrote: ↑Wed Apr 29, 2026 1:11 pm Thanks for that. I'll look into it. I was aware someone had brass gears made a few years back, and for a time sold them on ebay. They were $200 as I recall. When the HP 8640A began to play up because of the disintegrating gears, I bought a Marconi 2019 from Stewart of Reading and I like the push button control of everything, so there wasn't that much of an incentive to fix the HP8640A.

I think quality RF TE has always been relatively expensive. The lowest spec Siglent and Rigol models, which go to 2GHz and 1.5 GHz, are about £2,000 inc VAT. The top end models which go up to 40GHz, are well over £30,000. I haven't looked at Keysight, R&S etc prices, but I suspect they are a lot more.

https://www.ebay.com/itm/154054911606?_ ... R8CRqZ27Zw

I think this is the guy I bought a set from years ago. Haven't had to replace them (and at this point am not sure where they even are in the house), but they seemed to be decent quality. If the link doesn't work, just search "HP 8640 Gears" on the Bay of Evil.

-Pat

The "Curcuit" for this RF signal generator is suspect, the JFET Q1 doesn't appear to have anything but a capacitor connected to its drain.

Cubdriver wrote: ↑Wed Apr 29, 2026 3:24 pm https://www.ebay.com/itm/154054911606?_ ... R8CRqZ27Zw

I think this is the guy I bought a set from years ago. Haven't had to replace them (and at this point am not sure where they even are in the house), but they seemed to be decent quality. If the link doesn't work, just search "HP 8640 Gears" on the Bay of Evil.

-Pat

Thanks. I'll order some. They are a lot cheaper than the ones I saw a few years back.

synx508 wrote: ↑Wed Apr 29, 2026 3:59 pm The "Curcuit" for this RF signal generator is suspect, the JFET Q1 doesn't appear to have anything but a capacitor connected to its drain.

When I was looking this up on the web, there was some talk that the circuit around the FET was wrong in most downloadable manuals, but the Leader manual had an accurate circuit diagram.

To be honest I didn't look closely. It worked and I couldn't see much point messing about to improve it.