Call Search

New to Ham Radio?
My Profile

Friends Remembered
Survey Question

DX Cluster Spots

Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement

donate to eham
   Home   Help Search  
Pages: [1] 2 Next   Go Down
Author Topic: How fast do ALC loops react?  (Read 20827 times)

Posts: 67

« on: March 24, 2012, 05:26:47 AM »

Most external linear amplifiers make use of the associated transceivers ALC input to control the drive power and hence their protection.

Generally valve based linear amplifiers can withstand a high degree of abuse for a reasonable time (I guess seconds?) before permanent damage occurs however this is not the case with solid state external amplifiers.

From Radio Frequency Transistors – Principles and Practical Applications (Norm Dye & Helge Granberg) page 141

“Most transistor failures in solid state amplifiers occur at load mismatch phase angles presenting a high current mode of operation to the output transistor(s).”

“Since the temperature time constant of a typical RF power transistor die is 0,5 – 1,0ms, any protection system employed (including all delays in the AGC/ALC loop) must react faster than this time.”

It would be interesting to measure the external ALC response time of amateur transceivers – as I don’t have any such equipment it would be appreciated if someone with access to an oscilloscope could do this and report their findings.

This follows from a disappointing read of the negative comment about SS Amplifiers on eHam at “Ameritron ALS repairs”

I believe transisors and FETs are inherently far more reliable than valves provided they are operated within their specified limits.

Perhaps it’s the associated transceiver’s  ALC response time that’s  part of the problem together with the unnecessary need for SS broadband amplifier designers (driven no doubt by user expectations) to maintain a constant forward power into a mismatched load - sometimes up to a 2:1 SWR!

Designs that accommodate such user expectations can result in the final transistor/FET dissipation increasing by up to 70% for hardly 2dB more load power when operating into a mismatched load!

Consider 4 x MRF150’s expected to deliver 600W into a 50 ohm load – with a typical 0,5dB loss in the output circuit comprising, LP Filter, switches and directional coupler the 4 FETs must produce about 670W of RF.

Using a typical 45% efficiency this equates to around 200W dissipation per FET, this already limits the heat-sink temperature to about 50 deg C.

If the load now happens to be a SWR of 2:1 and forward power is maintained at 600W  the dissipation / FET can be >300W/FET exceeding the maximum specified dissipation for the MRF150 even at a junction temperature of 25 deg C!!

Clearly this is an undesirable situation for the poor MRF150.

Unfortunately with solid state amplifiers there's no cherry red anode or high grid current meter to alert the user – just a sudden reduction in output power and many wasted $$$ as we see on the forum.

What can be done?

Don’t deliver on user expectations or be priced to high due to conservative design and you’re out the market!

Reminds me of a rock and a hard place – what do you think?

« Last Edit: March 24, 2012, 05:30:34 AM by ZS6BIM » Logged

Posts: 933

« Reply #1 on: March 24, 2012, 06:12:42 AM »

"If the load now happens to be a SWR of 2:1 and forward power is maintained at 600W  the dissipation / FET can be >300W/FET exceeding the maximum specified dissipation for the MRF150 even at a junction temperature of 25 deg C!!"

This depends on the nature of your 2:1 mismatch. If the load has an impedance of 25 ohms then the transistors will draw excessive current, if the load is 100 ohms then the drain voltage will increase so that 600W can still be supplied.



Posts: 153

« Reply #2 on: March 24, 2012, 06:14:30 AM »

What paraemetrs are taken into account in defining reliability? Whether valves or solid state is more reliable depends on the criteria applied.

Posts: 7718

« Reply #3 on: March 24, 2012, 06:36:08 AM »

The 2:1 VSWR on the low side of the design impedance is the tough 2:1 VSWR impedance to work into.

Given an amp designed to just deliver X watts into 50 ohms, into 25 ohms the current is 1.4X greater and the voltage delivered is 0.7X as much. That means 1.4X more FET current with a greater voltage drop. For a class-B amp the DC-RF efficiency drops from 71% to 50% raising FET dissipation by 2.5X.

The increase in FET dissipation is not this dramatic for an amp that is not swinging to the rails to begin with. For a class-B amp that has a DC-RF efficiency of 50% into 50 ohms the efficiency is 36% into 25 ohms. The FET dissipation has increased by 1.8X.

What does this mean for the MRF-150 running 150 watts RF output? Into 50 ohms, at 50% efficiency, it must dissipate 150 watts. Into 25 ohms, at 36% efficiency, it must dissipate 267 watts. The MRF-150 thermal resistance (junction-to-case is 0.6 deg C/W).

So, into a 50 ohm load the junction runs 150W X 0.6 deg C/W = 90 deg above the heatsink temp. Into a 25 ohm load the junction runs 267W X 0.6 deg C/W = 160 deg above the heatsink temp. The maximum specified junction temp of the MRF-150 is 200 deg C. Exceeding this will not result in instant destruction; silicon can run around 300 deg C.

To run 267 watts and maintain the junction below 200 deg C requires a copper heat spreader and possibly a water cooled plate. The ALS amps don't use either of these. And how well are the FETs matched? The FET with the higher transconductance will hog current and run hotter. While paralleled FETs do tend toward equalizing their current it is not 100%. So, don't match the transconductance well and you have an extra hot FET.

It is junction temp that kills FETs. Overcurrent and overvoltage are just ways that lead to destructive junctions temps. Overcurrent obviously by increasing junction dissipation and overvoltage high enough to avalanche the FET to create localized heating of part of the junction.

Given that the Ameritron ALS amps are used in amateur service such as OOK CW, the dissipation numbers calculated above are cut in half. Now the dissipation is 75 watts into 50 ohms and 134 watts into 25 ohms. The resulting junction temperature rise over the heatsink temp is then 45 and 80 deg C respectively. That's not so bad.

Now back to the original question which I will paraphrase as "will overdrive, as when the ALC loop has not caught up with excessive RF drive, result in increased junction temperature?" The answer is no because as the amp is run harder the efficiency increases. As we saw from another Eham thread the dissipation is almost constant with RF output power. Drive the amp into flat toppping and the dissipation drops even further.

« Last Edit: March 24, 2012, 06:49:49 AM by WX7G » Logged

Posts: 67

« Reply #4 on: March 24, 2012, 06:40:38 AM »

This depends on the nature of your 2:1 mismatch.

Yup! you are right, dissipation depends upon the load at the drains that will have been rotated somewhat by the LP Filters, however as Dirty Harry asks - "feeling lucky"?  Wink

Second consideration - to achieve the 600W into 50 Ohms from 4 x 150W FET's requires that the amplifier be operating very close to voltage saturation; to now get 600W into 100 Ohms will require that the DC supply voltage be increased to 70V or do you just accept the severely degraded linearity?

What paraemetrs are taken into account in defining reliability? Whether valves or solid state is more reliable depends on the criteria applied.

I would say the wear-out mechanism of the device.

With power semiconductors I believe it's metal migration, very much a function of the junction temperature, however if the junction is kept below 200 deg C (the specified maximum) semiconductor wear out is not a problem in one lifetime Smiley

Regarding valves - I have heard of some low power valves lasting many decades however the higher power ones don't seem to do that well. Maybe 10,000 hours?
It would be interesting to hear from knowledgeable readers on this subject.



Posts: 67

« Reply #5 on: March 24, 2012, 06:55:28 AM »


Couple of points to consider.

600W out means at least a 670W amplifier.

Thermal resistance junction to heat-sink normally requires 0,15 deg C/W added to the 0,6 deg C/W.

It's unlikely an amplifiers thermal protection will come even close to tracking the junction temperature so the user will not know what is going on inside his FET.

I would not like to have a product with a component, that happens also to be the most critical and expensive, used way above it's specified limits.

Depending upon the out of band impedance presented to the amplifier's harmonics by the LP Filters it is not true that efficiency increases and dissipation decreases when an amplifier is over-driven.

Also consider that once the amplifiers output is no longer linear the NFB loop is effectively broken increasing gain and the possibility of instability.


« Last Edit: March 24, 2012, 07:09:39 AM by ZS6BIM » Logged

Posts: 7718

« Reply #6 on: March 24, 2012, 07:08:20 AM »


Ameritron does not operate the MRF-150 above any specification. Here are the MRF-150 maximum ratings and what Ameritron runs them at:

SPEC                          RATING          ALS
drain voltage             120 V            <100 V
Gate voltage               40V               <10V
drain current               16A               12.5A
Dissipation                300W              <120W

Don't bother picking apart my numbers by adding minor things I didn't include (for simplicity). I design high power electronics with my largest solid state linear amp being 100 kW.

The bottom line is that if the Ameritron ALS design is built correctly it does not abuse the FETs. But FETs do fail in these amps. So what's happening? I have to theories so far:

1) FETs not being matched
2) The bias circuit malfunctioning

Number 1 is based on Ameritron not selling matched FETs. Does this mean they don't match them when they build the amp? The specified transconductance range for the MRF-150 is 4 min, 7 typ, and there is no max spec. The EN-104 ap note upon which the ALS amps is based says to match the transconductance to within 10%.

Number 2, the bias circuit theory, is based on an experience I had with my ALS-1200. I shipped it and when it arrived one of the 50 volt power supplies would go into overcurrent and shut down. Readjusting the bias pots fixed the problem. So either one or more bias pots moved during shipping or another component in the bias circuit suffered a broken solder joint or a broken PCB trace.

« Last Edit: March 24, 2012, 07:25:11 AM by WX7G » Logged

Posts: 67

« Reply #7 on: March 24, 2012, 07:13:15 AM »

Don't bother picking apart my numbers by adding minor things I didn't include

What would that be David?


Posts: 7718

« Reply #8 on: March 24, 2012, 07:57:38 PM »

That would be the FET case to heatsink thermal resistance. What I'm saying is that I have refuted your theory that Ameritron operates the MRF-150 above the FET manufacturer's ratings. I don't see that you can fill the gap between your theory and my refutation with heatsink paste.

So, we must look elsewhere for the Ameritron ALS FET failures.

With my amp the incorrect bias didn't hurt the FETs as the PS would trip OFF when >25 amps was reached. What would kill (and did kill) a FET is having incorrect bias run a FET at 300 watts or above. I managed to do this while fooling with the bias adjustments. This is how I learn by breaking things. Another way to kill a FET is to have the bias jump up very quickly while the PS is ON. The stored energy in the PS output caps is more than enough to kill an MRF-150. So my focus is on the bias circuit and its possible misbehavior.

Posts: 92

« Reply #9 on: March 25, 2012, 12:55:41 AM »

That would be the FET case to heatsink thermal resistance...  fill the gap ... with heatsink paste.

Apologies for taking this quote entirely out of its original context; but it does raise an interesting question:

How well do manufacturers for the amateur market control the quantity, uniformity and thickness of the heat sink paste under each individual device? Or how well do you or I?

When we dismantle a CPU from its cooler, we see exactly how this should be done, with an extremely thin and perfectly uniform coating of heat transfer medium. Almost like paint, it seems to have been applied by screen printing or some similar process.

Also the interfaces are scrupulously clean, with not a speck of dust or grit that could act as an unwanted 'standoff'.

But that degree of control is only achievable in highly automated mass production. How can we (meaning you, I or a small-scale manufacturer) achieve that same degree of control in a hand assembled amp?

73 from Ian GM3SEK


Posts: 1071

« Reply #10 on: March 25, 2012, 03:09:39 AM »

Any designer that relies on an ALC loop to the driving transceiver to protect their linear amplifier, is foolish at best. There are so many variables in this approach, even if used with their own design of transceiver it relies on the user to connect the cable.

The amplifier should self protect, even when over driven.

Trying to use ALC as a means of power control creates an unknown ALC feedback loop in many circumstances, potentially creating other problems such as over drive and splatter.

Many modern amplifiers, such as the Acom valve amplifiers and most commercial solid state amplifiers, have built in self protect which switches an input attenuator or in extreme situations bypasses the amplifier completely.

73 Dave

Posts: 531

« Reply #11 on: March 25, 2012, 09:47:56 AM »

But that degree of control is only achievable in highly automated mass production. How can we (meaning you, I or a small-scale manufacturer) achieve that same degree of control in a hand assembled amp?
The bigger issue is that of heatsink flatness, which is IMHO honoured more in the breech then the observation by many manufacturers.
Sure, grinding and lapping the heat sink are boring, but they probably make more difference to the thermal resistance then any paste.

One trick that can work is to make up a simple jig with a bit of shim stock then use a credit card as a squeegee to apply the paste. Some of the modern phase change materials are potentially interesting, for all that they need the clamp bolts re tightening after operating at temperature.

Something else sometimes 'value engineered' out is the correct washers and correct torque settings, none of which helps, particularly if someone has gotten a little carried away with the paste dispenser.

Regards, Dan.

Posts: 9930

« Reply #12 on: March 25, 2012, 10:28:54 AM »

I never use alc  I prefer to set the power level my self.  ALC line never gets hooked up. If I blow it up then I pay the price, but I take responsibility for it myelf.

Posts: 815

« Reply #13 on: March 25, 2012, 01:17:03 PM »

On the subject of ALC use, many take it in different views without considering the benefits..
It is not meant to protect as such but as a dynamic limit on drive to both the amplifier and the driver as a combination.
Different model radios can handle ALC differently.
If one depends on mike gain only to limit output, there s no guarrenty that loud audio sounds won't drive the amplifier too hard on those peaks and cause QRM..
On a Kenwood TS480 there are a number of different circuits that put a lot of control in the transceiver before the drive ever gets to the amplifier.
Using ALC just adds to and enhances the whole feed back control system.
Also an element of RF compression is added to the signal and that does not mean distortion on splatter in this case since if the transceiver is not over driven, the amplifier is not overdriven with ALC feedback applied.
It all stays clean.
I use ALC on an AL80B full time and use it for low power inital tune and adjust average power out as needed after full tune.
This allows full transceiver output that is throttled by the ALC action.
By using transceiver full out , the transceiver has it's full range for ALC control action.
On AM mode,  I use the amplifier ALC control to limit the carrier power to about 100 watts even with full 25 watts out of the transceiver.
This allows for 400 watts PEP from a 1kw amplifier. Pretty neet to have for a 25 watt radio and run the 3-500 at about rated plate dissapation.
On top of all this I can even run speech processing if needed. It's still all under ALC control.
On AM you must watch the plate dissipation over time with long winded key up so the tube does not overheat excessively.
I have been running the setup for several years with no issues.
Since I run 5 bands out of an 8 band audio processor and Heil mikes, I care about station performance and sound as well, and never get  a report of splatter or distortion.
If it's all largely understood and setup correctly it all works.
Some of the things to know is amplifier tune and 'overloading' about 10%, and having the antenna tuner set up in it's sweet spot for overall match impedance (not just any combination that results in an apparent match). The tuner optimum setting was found with a Palomar Noise Bridge and the amplifier does like it. R=50,  Z= near zero not some value well into Z/c or Z/l .
There are other specific benefits but will let that lay for now.
Bottom line is these actions are not just plug and play or not used at all depending on where one stands on the matter.
ALC is made available as a benefit, if one understands it's dynamic uses.
Some radios do better with it, some not so well.
BTW, the 480 has little to no spiking on key up as I use it to drive 2m and 70cm solid state amplifiers as well as a 4cx400 6m amplifier..
Good luck.

Posts: 5688

« Reply #14 on: March 25, 2012, 03:03:53 PM »

I'm surprised that noone has brought up ALC Overshoot probs...

Pages: [1] 2 Next   Go Up
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!