Golborne Vintage Radio

Hi.

I've been trying to make a good quality pantry transmitter to feed my radio sets. I had set on using short wave and tried many approaches, all worked well enough but some distorted at 70% modulation or would FM the carrier.
When looking for an IC for another member here I found a drawer full of the ubiquitous MC1496 is P and N guise. I look at the datasheet I decided to have a go with the chip and a 15.060Mhz crystal. After many attempts using veroboard in the past I was reluctant to try it as I've been told in the past it's very lossy at RF.
A new LCR meter that arrived a few days ago was used to test the capacitance between adjacent tracks which only read between 2 & 5pf so I thought I'd give it a go. I made up an oscillator first using the crystal and a 2N2222A. The result was good with 500mv output from the 15.060Mhz Xtal. Next was to make up the modulator. The layout follows the circuit loosely and again I was rewarded with good results, some changes were made to the datasheet circuit, I replaced the gain setting resistor between pins 2 & 3 with a 390 ohm for more gain, I also added an RF transformer using toroid construction as loading pin 6 or 12 produced distortion. Results at this stage were very good. Adding any more than a metre length of wire for an antenna gradually brought in distortion and range was not good, in fact directly feeding a radio wasn't great either. So obviously an RF buffer was called for, initially I tried a bipolar buffer in different configurations, they worked but the waveform had deteriorated, things improved using a FET in the first stage but It still wasn't as good as I hoped. Next I tried a 1J18B which was a total cure. The only disadvantage of this approach is you need a high voltage source for the valve. The results are first class and I have to say I'm really pleased with the way this works. Signal level with a few feet of wire is good with range around 15 meters It can just be heard 30 meters away on a sensitive radio but its very weak, reducing the HT to the Rod Pentode reduces range if needs be.
On my circuit C1 is 270pf, C2, 100pf, C3 27pf, C4 22pf, Cx 56 pf at 15mhz but as much as 22nf on long wave.

Feel free anyone to use this circuit or share but I've already had my circuits appear elsewhere and whoever has done this has accredited the design to themselves which is rather annoying, if sharing please ask.

https://youtu.be/e2-fON-QTYQ

Interesting hybrid design. The 1496 is a very common part. It's certainly useful up to more than 50MHz, possibly to 100MHz. Works well in Band I TV modulators, using one each for sound and vision.

Thanks Jeffrey.
The 1J18B Screen is drawn wrong with it being fed with a 33k from HT & the decopler is 4.7nf.
The cap values that are not marked on the diagram vary with frequency used.

Corrected diagram below.

One thing that strikes me in the design is the way the output of the 1496 is connected. The usual arrangement for RF is a centre tapped transformer with the CT to +12V. This would probably give rather better drive to a transistor output stage.

Looking at the PDF datasheet, the output is VO+ & VO-
They are in opposite phase. In fact if you scope the waveform from pins 6 & 12 the resulting output is double. You can use a centre tapped tx and use the centre tap as the supply..... But it needs to be decoupled and distortion is huge. In the data sheet for various applications I see no centre tapped tx. I tried feeding a transistor buffer in many ways but the end result seems to be best for me.

I will though try it again with the centre tap direct to supply, as I fed it via a 1.6k resistor decoupled by 4.7nf.

This is the Rediffusion HF modulator which I repurposed for Channel 1 TV. One each for sound and vision: http://www.borinsky.plus.com/simple_channel.html It drives a 75R load with no problems. I can't remember the exact design of the transformer but I'm sure it wasn't critical. I've used similar circuits in PAL/NTSC coders.

You definitely don't want a big resistor in series with the centre tap. The outputs of a 1496 are essentially current sources. You're trying to develop a voltage across the load, not across a collector resistor. Actually a resistor to the CT isn't so bad, in theory there's no signal voltage across it. All it does is make the bias conditions more critical since it's effectively lowering the supply voltage. Remember that the current source at the bottom, signal input at the long tail pair, carrier input at the quad and output at the collectors mustn't clash, even at maximum swings.

Hi.
An early start this morning. I decided to retry the centre tap in the toroid primary but this time directly to 12v +.  This dropped the resultant output on the secondary by 15%. Trying a series resistor on the feed and decoupling the centre tap again brought in distortion and lower level, around 300 ohms and lower the distortion cleared up. I've tried another chip just in case the one I was using was faulty but no. So either it's an effect of the vero or the 3.3k's as pins 6 & 12 are in fact the optimal values stated by the manufacturer in the data sheet. I'll leave it as is I think.

3K3 collector loads may be the best value for maximum output swing but it's not a good way to make a modulator. The output impedance will approximate to 3K3 which isn't ideal.

Looking back at my own 1496 based designs. The 1st snippet is from a burst demodulator so it has a low frequency output. Hence a CT transformer isn't useful. I've already put my Band I modulator in a previous post. The 2nd snippet is from a PAL coder I designed many years ago. I re-used the basic design many times in later designs. I'm not sure why I bothered with decoupling on the CT, possibly to ensure that any unbalanced HF current was kept out of the main supply rail.

Thanks Jeffery.
The link to your article has the RF transformer configuration just about the same as I tried.
So can you explain why when I do it this way the output is lower level, directly fed its still clean with zero distortion but as I say lower level?

No I can't explain what's happening in your tests.

When you're working with a CT transformer the output will be determined by the current swing in the 1496 multiplied by the transformer ratio multiplied by the load on the secondary. If that's sufficient then all well and good. Otherwise you need gain afterwards. The voltage swing across the load will be reflected back to the 1496 outputs via the transformer. At RF you obviously want a fairly low load resistance. From memory the Band I modulator delivers about 50mV (probably p-p, not RMS) into 75R but it could be a little more.