Troubleshooting Hedghog

[peter scott]

Hi Nick

If I type "Hedghog 405" into Google it auto corrects to "Hedgehog 405". If I then press the "Search instead for Hedghog 405" button all the stuff about the converter comes up in the results.

Farnk

Hi Frank,

My natural inclination would be to spell it with an e. I may be a bit thick but is there a subtlety to the "e"less spelling?

Peter

[Murphyv310]

Both the Raspberry Pi & Hack RF can do modulation but and its a big but can they do non intercarrier systems like System A ?

[ppppenguin]

Peter, Frank told me about the missing "e" when I helped with his forthcoming Bulletin article. All will be reveealed then, unless Frank wants to say something here.

Trevor, at the price of R-Pi, if they can't do intercarrier sound then just use 2, one for audio, one for video. There are 2 modulator chips in the Aurora and Hedghog simply because one couldn't quite manage AM sound at 3.5MHz spacing.

It might be possible to do intercarrier sound with System A but it's difficult with any +ve modulation system. Minimum vision carrier must not fall below a certain amount, otherwise sound will be clipped. Since the sound is 6 or 7dB down wrt vision this would mean doing both of:

1: Placing sync tip at something close to 50% carrier. It should work but might fool some sets
2: Reducing the sound carrier level, which would reduce the need to raise sync tip. Again this would work but might annoy some sets.

Yes, you could do iintecarrier System A after a fashion but it would be horrible. It may well be the case that the R-Pi methods can't generate 2 separate RF carriers from a signle computer so that menas using 2 of them.

[Murphyv310]

Then it gets really messy Jeffrey.
So all you have is a dedicated modulator like the David Looser one and the Darius, not a huge issue as the original converters didn't have a modulator. The David Looser one works fine on all band 1 channels and could even be switchable crystal controlled if needs be.

[ppppenguin]

Since custom crystals are expensive there's a possible trick using the FPGA to synthesis the carrier frequencies for conventional modulators. There will be phase jitter from the discrete time oscillators but I don't know how much it would matter. Depending on the PLLs availabale on the chosen FPGA it may be possible to directly synthesise something close enough to the carrier frequencies with minimal phase jitter.

For those not familiar with DTOs in their simplest form they are just adders and accumulators. On each clock cycle you add the coefficient. So the output frequency is:

Fout = Fclock * Coefft/2^n

Where n is the number of bits in the accumulator. Maximum output frequency is half the input. The output is the adder overflow. More complex arrangements can be used to give ratios where the denominator isn't a power of 2. The accumulator output can also be sent to a sine table and DAC to synthesise a sine wave output. Typical numbers of bits range from 16 to 32 or more. Very simple to implement in a FPGA.

[`FRANK.C]

Hi Peter
No, it was just me been thick  "Hedghog" comes from a spelling mistake I made on the PCB silkscreen and I then decided to keep the name. It has given me a chuckle or two as it does raise a bit of curiosity. It has an unintended advantage in that when spelt "Hedghog" in google it helps to cut out the unrelated "Hedgehog" stuff.

Among other things I hope someday to revisit trying to do the modulation within the FPGA. The last time I tried I felt that the problem was the R2R DAC wasn't capable of handling the high frequencies.  Next time I will use a AD9744 DAC chip which is a 14 bit device capable of up to 210 MSPS. At best it will be only able to cover Band I. I want to try and synthesis a sine wave carrier using a DTO and look up table. As far as I understand the sample frequency will need to be at least 2.5 times the carrier. Which will make the sample frequency close to the limit of that DAC. I expect there will be plenty of spanners ready to fall into the works. If I get around to trying that I will start a new thread.

Frank

[ppppenguin]

The vision modulator isn't too complex in theory.

Use a high frequency clock that's a multiple of the 405 line video clock. This means you don't have to get a signal between async clock domains.
Generate the carrier frequency with a DTO.
Interpolate the 405 video up to the HF clock rate
Multiply the video by the carrier
Send the result to the DAC

None of this is too hard, except that as you push to much over 100MHz the FPGA will start to complain. Then you'll be using all the tricks in the book to get the relevant parts to work fast enough. Been there, done that., it's not pretty. Newer, faster FPGAs will be better. To give you an idea of improvements over the years, when I first used FPGAs in the early 1990s I was wary of running at 27MHz and so used pairs of 13.5MHz video streams. This may have been overcautious and made me use more or bigger devices but it made life easy. With Xilinx Spartan 3 series FPGAs (as used in Aurora, Frank's Altera in the Hedghog is of similar vintage) I've found that 54MHz is utterly uncritical, 74.25MHz is very easy but 148.5MHz needs great care. Any higher would get horrible, except perhaps for small amounts of logic.

The sound modulator is in some ways more of a nuisance. You'll need an audio ADC, driven by a clock that's an integer fraction of the HF clock. Then it's much the same as video. The DAC may or may not be sufficiently linear to convert video and audio without bad sound on vision or vision on sound. If not, then you need a 2nd DAC for audio.

[`FRANK.C]

Hi Jeffrey
I will be using a Cyclone IV EP4CE10 on a development board made by Waveshare. It has 2 mm headers which are shorter than the 2.54 mm ones. I have a PCB made up with most of the components mounted on it. The ADC is a AD7476 which is a 12 bit that the sample rate can be controlled by the FPGA.
It's all a long way down the track yet.

Frank

[ppppenguin]

Glad you're using a more powerful FPGA. I've spent far too much time struggling with various FPGA limitations. Whether size or speed related. If you're developing a cost sensitive design it may be worth jamming the logic in the cheapest possible FPGA but otherwise it's a particular corner of hell.

The AD7476 seems a strange choice for audio. There are loads of 16+ bit ADCs designed specifically for audio that will give optimum results with minimum effort. I've never had to design kit that needs an audio ADC so can't make any specific recommendations. If the AD7476 is already on the evaluation board then it will work for audio but 12 bits aren't really enough for high quality.

[peter scott]

Hi Peter
No, it was just me been thick  "Hedghog" comes from a spelling mistake I made on the PCB silkscreen and I then decided to keep the name. It has given me a chuckle or two as it does raise a bit of curiosity. It has an unintended advantage in that when spelt "Hedghog" in google it helps to cut out the unrelated "Hedgehog" stuff.

Frank

Thanks Frank, and I like the unintended advantage.

Peter