Dummy load is 50 ohms resistive, my electrically rather too short vertical is nothing like that, and setting the transmatch does require running a **Little** power (Roughly 1W to get an acceptable return loss measurement with my meter) into the thing so I can actually you know, find the null on the reflected power meter.......

'Tuning up is prehaps a misnomer in this context, I have a perfectly good LO derived from an OCXO, tuning up is a very occasional activity involving a low Allan variance source locked to GPS reference and measuring to fractional Hz precision the 100Mhz OCXO  then telling the LO generators control micro about the measured value, it does not involve radiating much of anything.

Dummy loads are wonderful things and should certainly be used when  working on the rig, but the applicability to working on aerials and their associated matching networks is marginal (yea, I am lazy, I am not going to tune the AMU with a vector impedance bridge, just not going to happen).

Regards, Dan 2E0CHE.

OUCH!
That is a nasty bug.

Regards, Dan.

Or possibly the guys you hear tuning up to top of the DX actually CANNOT HEAR the DX or the person working them if they are also DX (Plenty of us out there with inner city locations and S6 QRM all the time)!

This is why my standard tuneup is "M6ATV[1] tuning up, is this frequency in use?" Then LISTEN, then go brick on key while I fiddle with whatever needs fiddling with, then "M6ATV[1] tuneup complete" then Listen again to see if any apologies are in order.  For the most part tuneup can be done at well under a watt if your instruments are of reasonable quality, but there are some things you  need to do at real power.

I would note that this tuneup is actually illegal as it is neither a CQ call nor addressed to a particular callsign, but it seems to me that common sense almost certainly applies here. 

I don't DX (No point in getting involved with my QTH), but I would be mortified to find I had stepped on some ongoing QSO where either side was even marginally audible to me.. (If I cannot hear either side, then well sorry, but how am I supposed to know what is going on?).

One other gotcha is that (PSK guys looking at you) if your signal is 31Hz wide, then you really should be using NARROW IF filters, if I am 2Khz away I may not even know you are there is I have 500Hz or 250Hz quartz in the radio and a good frontend, no point bitching because you are using a 2.7Khz wide SSB filter for a narrowband mode and I have just wondered into your over wide IF passband with a 65db stronger signal.

Deliberate jamming (even of something you don't care for) is just wrong, turn the dial and find something you are interested in, there is plenty of space.

Regards, Dan (M6ATV/2E0CHE)

[1] Now 2E0CHE, passed the exam.

Assuming we are talking AGC loop induced overshoot then yes, you can damage a FET amp that way, but really you should be setting the drive power so the AGC loop at most barely activates.

If you Have say a 1KW amp with a known gain of say 17db (so it needs 20W of drive for full output), then provided you set the rig to output 20W maximum the AGC loop will never activate and thus never overshoot......

If the rig is overshooting even without the AGC loop involved, then the answer is probably to run the rig at full power and just stick a suitable pad between the rig and the amp (This will also improve the broadband match seen by the rig as well as the input match seen by the PA).

Regards, Dan.

I might agree about the ADT-2000A, but IMHO the FTDX5000 takes a rather brute force approach and still does not get anywhere near what is possible (Hint guys, 75W of class A is not a selling point in my book, the shack is warm enough without that).

I have an experimental homebrew rig that manages better then -60db third order TX IMD ref one tone, (Equates to -66db ref PEP) at several hundred watts output with PA efficiency of ~50%, it is not that hard to do, and before I am done I hope to push the efficiency up quite a bit.

Transmitter performance these days in amateur service is by and large pathetic, at least compared to what is possible, or even the minimum that would be acceptable in commercial service.

Regards, Dan.

Two tones spaced closely in frequency are mathematically identical to a single tone half way between the two, DSB modulated by half the difference frequency.

sin (a) + sin (b) = 2 sin ((a+b)/2) cos ((a-b)/2) going from very rusty memories of the relevant trig identities.

Thus a close in two tone test can indeed reveal 'issues' with both linearity and the gain distribution.
It should not be an issue with a well designed modern rig, but sometimes still is.

Agreed about the notched noise test, a good check for LO phase noise and reciprocal mixing, that even some very expensive radios get into trouble with (and not just SDRs).

Regards, Dan.

Sounds like its really just testing the same thing a two tone IMD test does?

Regards, Dan.

Forget audio amplifiers, the ONLY similarity between audio and RF power stages is that both tend to use impure sand as an amplifying medium....

Look an audio power stage typically has open loop gain for days, and applies huge negative feedback around the whole mess to linearize itself, an RF stage does not have that option as the phase shifts through multiple stages will make the loop impossible to stabilize.

About the best that can be done is to feed the envelope back (and even this is not trivial), and in practice most RF amps run almost open loop (Apart from the intrinsic feedback from emitter impedances and miller capacitance and such), at least the budget stuff for amateur and most commercial service does!

Now consider that the Vas and input stage in a standard audio amp will be running from a regulated supply in all probability, it is only the current amplifier stage that is actually unregulated and that these things use voltage feedback.....

In the case of your T1000, there is sufficient capacitance in the power supply to effectively bypass the audio bandwidth envelope current (at least mostly), so while it is not ideal, it does save on a high power DC regulator and its associated heat making it a reasonable tradeoff as long as the amp still met its design specifications. 

Regards, Dan.

You are looking at a GaAs device for microwave operation there (Probably for the mobile phone or CATV market), power probably means a hundred mW to a watt or so, rather then the sort of silicon based thing operating at a 20dB higher power level used in your HF amp.

Different technology in a different operating region.

You really do need a copy of either EMRFD or Dye and Granburg......

Regards, Dan.

Ze in a RF power device is usually MUCH lower then 1 - 10 ohms! Consider what that would do for a device running say 100W on a 24V supply rail, so a peak current of around 8A.

A 1 ohm Ze would drop 8V, a full third of the supply voltage!

Consider also that that the base impedance is typically low (~10 ohms on a good day), and thus that the base voltage swing is small for a given drive power.... Re wants to be made as small as possible of you will loose all the power gain to negative feedback.

There is a reason that RF power devices come in funny packages with two or more wide low impedance emitter/source connections, it is that Ze must be kept very low.

Seriously you want a copy of Dye and Granburg for the gory detail.

Regards, Dan.

The emitter resistance does (as does the intrisic emitter resistance of any transistor) of course reduce the gain via negative feedback, but that is NOT a desirable thing and great care is taken to make the emitter connection have an impedance as low as is possible while providing adequate control of any hotspotting.

The key thing to realize is that unlike in the audio case, these transistors are usually operating close to Ft and usually have very limited power gain (10 - 20dB typically), which inherently limits the ability to apply negative feedback as it robs gain.

When a stage only has a power gain of 13dB open loop, feedback to improve linearity gets tricky, and feedback around multiple stages tends to lead to instability as the phase shift increases.

RF is not audio, and the design approaches are very different. For example one trick (more useful in a rig then an external PA) is to apply envelope feedback around a few stages, this does nothing for the shape of the individual rf cycles, but can reduce the IMD by reducing the envelope distortion. The low pass filters are of course still needed to remove the harmonics due to the distortion of the individual cycles. You don't see this much in ham gear, but it is popular in commercial equipment.

Regards, Dan.

One other thing to watch is that phase noise does not tell the whole story, an SSA set to measure PN will in some modes ignore amplitude noise which can be just as big a problem.

See Martein Bakkers work on the AD9910 for some unpleasant plots due to a simple decoupling screwup by AD (Unbypassed on chip bandgap), excellent phase noise performance but the AM noise is painful.

Like all engineering specifications, phase noise data often needs a little qualification and should never be taken as a figure of merit without also considering how it impacts the rest of the radio.

Regards, Dan.

External PAs  are tricky things!

A semiconductor external amp can usually be brought into the loop without too much trouble (My experiments are using an old military amp brick (BLW50F *4 and BLW96 *4 for ~500W, a blast from the past!), but that is because its bandwidth is much greater then the bandwidth of the correction loop. The same cannot unfortunately be said for a typical tube PA with its high Q tank circuit.

Due to the fact that until I take the next level exam (a few days) I am restricted on the air to commercial kit all my experiments have been into a dummy load (Hate the UK licensing setup - why can I not just take the advanced exam? Been designing commercial transmission gear for years).

Basically my loop looks like this:
There are two DDS chips fed by a common clock, one driving the forward direction switching I/Q modulator and the other the mixer for the reverse path from a power sampler at the output of the PA. To compensate the group delay through the PA and so bring the forward and reverse path baseband I/Q pairs into approximate phase so they can be subtracted, the reverse path DDS is programmed to a hundred Hz below the frequency of the forward path, so that the phase of the received IQ pair slips relative to the transmitted pair, the loop is only closed once the point corresponding to minimum error term is found.

A small micro (ARM LPC17xx) controls the whole mess.
There is actually a third DDS to take the IF up and down from the transmit frequency, but that is a detail.

So far results are promising.

I dislike RF AGC in PAs, the approach is just wrong!
You know what the PA is rated for and measure its gain on each band,  so you know the required drive power on each band..... The rest is just a case of setting the drive power limit appropriately. The loop bandwidth inherent in a typical PA AGC loop is too small and as you never really know the transfer function of either the rig or the PA, so how is that loop supposed to work correctly?

If you sold the things, some 'screwdriver expert' would find a way to screw it up, probably by monkeying with the power detector head. See enough of that even on broadcast sites.

Given that the gear manufacturers mostly don't even do the basics when it comes to lowering two tone IMD (Diplexer filters are not rocket science, and they help), instead preferring to boast about how inefficient the rigs are "Class A to 75 watts!", as if that was a selling point, I don't see  commercial predistortion or Cartesian loop (not quite the same thing!) appearing in the mainstream any time soon.

Regards, Dan. 

Getting vaguely back on topic, there is more to phase noise then just the oscillator in use.

Power supply noise, spurs picked up from bad PCB layout, any divider chip (Say for I/Q generation for a mixer) can add noise and of course the ADC will have aperture jitter.....

at below about -150dBc you need to pay massive attention to all of these to get a good result, so phase noise differences between radios using the same clock generator are to be expected. 

This incidentally is why band limiting the RF before the mixer/converter helps (for all it goes against the grain when doing SDR) as it reduces the total noise power for the phase noise to reciprocally mix with. 
 
Ethernet has the potential to be a big win, but for my use case the price would need to come down first, the early stuff is always expensive.

Regards, Dan.

Sort of.... Splatter is usually seen as IMD causing totally excessive transmitted bandwidth (Usually caused by clipping something), by definition the only things that can cause received audio problems are things that fall within the receivers IF passband (So some IMD counts), where splatter is normally relatively close but outside the receivers passband (so the receiver may be unaware of it), it poses a problem for other users of the band trying to work close to the frequency of the problematic station.

For example if the receiver has a 2.4Khz SSB filter then only IMD falling within that 2.4K bandwidth will be a problem for the receiver but if the transmitter is generating sidebands full of splatter out to 10s of Khz or so either side of its nominal frequency then someone trying to work DX 5K up the band will be cursing you.

Also be a little careful of thinking too strictly in terms of bands, the RF really does not care, it is perfectly possible to splatter over a band edge.
Be a little careful here, spurs and harmonics cover slightly different cases.
A harmonic is an integer multiple of the input frequency, a spur (while it may be a harmonic) does not have to be an integer multiple. For example some RF amps can suffer from low frequency instability in the audio or ultrasonic region at some drive levels,  which can cause masses of spurs that are not harmonically related to the output (They are in effect modulation sidebands) sometimes extending Mhz from the transmitted frequency.

The T1000 looks to be basically a datasheet implementation of a push pull PA of a fairly crude sort, and like all amps will have IMD and harmonic generation issues, the key thing is to understand what the filters can remove and what they cannot.

If you want to dig in more depth then the following are good references:
"Radio Frequency Transistors", Dye and Granburg - They wrote most of the Motorola app notes for RF transistors.
"Experimental Methods in RF Design", ARRL, excellent.
"The ARRL Handbook", ARRL, much good discussion on this and much else that any Ham inclined to technical fiddling should know.

Regards, Dan. M6ATV.