500 Watt PA

[VK5YYY]

I scored a UHF TV transmitter from a site that was upgraded to digital.
The unit is a 500W Teko on Australian ch 32 circa late 1990 and may be suitable for our local repeater on 446.5Mhz.

With regard to filters when I check the local TV the signals are very defined SkyScrapers in a row.

1 What style of filtering do commerical digital TV PAs use and where is it located, are there any designs floating around?
Guessing interdigital high power filters are used between PA and antenna...maybe something between stages.

2 What do I do for a 1 KiloWatt dummy load, designs please....smaller loads mounted on a big heat sink with a splitter network maybe?

3 Power measurement, is buying a proper digital Bird meter the way to go or is there a cheaper way, maybe calibrate an analogue Bird meter?
Is there a rule of thumb between analogue and digital measurements, the unit has metering via couplers, how do I calibrate it?

Any advice greatfully accepted

Cheers Roger VK5YYY

[VK5YYY]

We tuned up the Combline output filter on the PA tonight
The unit was Ch32 AU 594Mhz and it came down to 446.5 no probs
Input loss was 1.3db which is the same as the as found loss (some of which is connectors and cables)
It's the first filter I've seen with a heat sink
All these TV filters have "egg beater" coupling adjustments, more coupling wider band width
There is also 2 pods on each end that look to be notch filters
If someone could explain why, that would be good......intermod?

[VK5YYY]

From last few days reading- a 500W analogue transmitter has to be derated 6db to run digital COFDM
So my 500Watts is now a measley 125Watts : (

I think I have most of the answers to my original questions (incredible really: over 500 reads of my post and no responses)

1 Filters do nothing to improve the signal to noise of the recieved signal, they only protect the guy with the next door frequency
from you Noise + Intermodulation = Spectral Regrowth. Generally located at the output of the final PA and dump another 1db or so off your signal

2 Large UHF dummy loads are not available in most Ham Sheds, but 125Watts I think I can handle

3 Using a spectrum analyser coupled to the load at resolution band width 300khz the measured power is around 10db down from the actual peak
(Until Bevan discovered the channel power button on the Devisor, too scared to connect the Fieldfox)
DVB-T power measurements with a Bird meter I'm guessing are not real definitive and might point to RMS or Average power

Could one of you wise gentlemen set me straight on that stuff if I'm wrong

[g8cpf]

Lack of responses might indicate the VERY select area of engineering in which you are exploring .... Very few of us will have had the privilege of working on such equipment..... Well done with your research... not easy to find such specialised info these days.

[VK5YYY]

Oh....Well I'll be sure to fill you in on what happens next, we'll be firing up one half of the Txer this week....Smoke maybe?
If anyone wants any bed time reading one of the best texts is from these guys that wrote reports prior to the introduction of DVB-T in VK
http://happy.emu.id.au/lab/
I'm guessing these guys had some well paid fun telling the broadcasters and government how to go digital circa 2000
Most of them have amateur back ground too
Governments love selling spectrum so these chaps drew up the map to the money tree

[VK5YYY]

Well, I have attained some expert advice very greatfully recieved from Neil Pickford:

"to get the linearity you need to back off the digital PA power 6-10 dB from the peak vision power used with Analog TV, although DTV can survive a bit of non linearity since there are pilot carriers inserted every 4 symbols that are used by the receiver to equalise the channel before/during demodulation.

As to Power - The IFR (spectrum analyser) will will give you a relative power measurement but at 300 kHz RBW you won't be capturing all the power in the channel. The only true way to measure DTV power (average) is with a bolometer type power meter (this measures the heat dissipated into a load) as your friend said "boil water". Bolometers are available to measure power levels from 10s of Watts right down to nW using a crystal bolometer.

Your other low power meters can be adapted using calibrated high power attenuators (I use a 40 dB 50W Narda) or a calibrated/terminated directional coupler.

Your Bird meter is a peak voltage reading device that has been scaled to indicate power based on assumed currents in the line and these don't work very well on DVB-T as they are not calibrated for a COFDM signal. There is no such thing as RMS power. Power is always Average power. RMS is a conversion used with sinewaves.
For a CW (sinewave) signal the Peak voltage (scaled to RMS) x Peak current (scaled to RMS) = average power, however OFDM is NOT a CW signal.

The ACMA definition of Px is:
Px - the average power delivered by a transmitter during one radio frequency cycle at the crest of the modulation envelope taken under normal operating conditions. This is commonly referred to in amateur circles as peak envelope power (PEP)."

So I dug out the old IFR6960B power meter that I haven't used since I got a decent spectrum analyser and fired it up
I just goes to show sometimes the old things are the good things

[PHYguy]

Re. Measuring DATV power with (traditional, swept RF, analog IF) spectrum analyzers, have a look at:

Agilent Technologies, "Measuring Noise and Noise-like Digital Communications Signals with Spectrum and Signal Analyzers"
http://cp.literature.agilent.com/litweb ... -4008E.pdf

After all the math, you'll find you may need to add ~2 dB to the SA readout and an additional factor (converted to dB) that represents how much smaller the SA resolution bandwidth filter (RBW) is as compared to the overall bandwidth of the signal. (E.g. 14.3 dB using a 300 kHz RBW on an 8MHz signal.)

Also use SA video bandwidth (VBW) at least 10x smaller than RBW and take care not to overload the SA mixer. The displayed signal level looks lower due to the smaller RBW, but you're still driving it with all the power in the signal BW. (And might end up measuring the SA mixer's non-linearity instead of that of the PA.)

Regards,
PE1GTA

[VK5YYY]

We built a test exciter and that pumps out a watt or two of DVB-T
This consists of a commercial Antiference DVB-T modulator into a HA-40125 Hills (Polytron) 40db MATV Amp, that feeds a 20db ABE push-pull 20W PA
Provided you don't drive things too hard the output is very clean, the Modulator has a MER around 30 which degrades to about 29 through the chain

The key is to drive everything 6-10db below the analogue rating, beyond that spectral regrowth takes over and MER drops rapidly
It's a bit strange having all these very powerful amps in sequence only to generate a small amount of power "MORE IS LESS"

We now have a bolometer style wattmeter (although we found the Bird meter measures within around 2db of the true power)
The Bird meter jumps around but the Bolometer is very stable
TV transmitters have "interlocks" which prevent things working under fault conditions and allow things to power up in a sequence
A nice man at Broadcast Australia told us how the bridge the interlocks so we could power things up
By chance a broken wire was found on the output power measurement coupler so we can now run the whole unit properly
We've tested it to around the 125W it's quite clean

[VK5YYY]

We now need a decent load to test the PA into so I bought some used 50 Ohm 250W Anaren loads from Ebay they were about $4AUD each (10 buy)
I bolted 8 of them to an old inverter heat sink into 2 groups series/parallel combination to 2 n types the VSWR was shocking
Great idea Danny....whoever listens to radio mechanics anyway

Then I tried 4 individual loads each with it's own n type, I used an old RFI cell phone 4 way power divider to each device the VSWR was much better
but the spitters VSWR went bad at around 600mhz and at 70cms the VSWR was way to high
The power divider is just a lenght of the correct impedance coax that works great at 850-1800mhz but not at 70cms

Then we tried an old Rojone cell phone 4 way strip line wilkinson style power splitter the VSWR was around 1.5 and with jiggling I got it down to 1.1 PERFECT! Only tested it with 25 watts on the bench but looks very promising, testing on PA tomorrow
Tidy up the cables a bit and maybe add a fan for good measure, what's the worst that could happen......Smoke?

[VK5YYY]

We've made the thing a little more compact and have and old aluminium ballast box from a sodium light for a housing
A 240vac Papst fan runs a one end and creates a fairly major draft
Notice the 2 way power splitter in the background does not have parallel inner it's stepped every 40mm,
that's their way of transforming it into a broadband device