Modifications to Increase the Efficiency of the Ramsey AM-25 Transmitter

radio8z's picture

The body of this report is posted as a PDF attachment in order to preserve the formatting.

Mar 7 This post has been edited to contain new information.
Mar 17 An error was found on the schematic for the modified output stage. The values for C3, C4, and C5 were given in nanofarads. The correct units are picofarads. The schematic has been corrected and is marked as Rev 1.


Ramsey AM-25 Efficiency Enhancement

Increasing the field strength of a radiated signal will, in general, increase the usable range of the signal with the range increasing approximately as the square root of the output power. Since the DC input to a Part 15 AM transmitter is limited to 100 mW under the Part 15 rules (15.219) increasing the power out is not simply a matter of increasing the power in. Another approach is to improve the efficiency of the final stage so the output power is increased without exceeding the 100 mW input power limit. Changes to the Ramsey AM-25 are described here which resulted in more than double the original circuit efficiency and which thereby doubles the power output while maintaining compliance with Part 15.219.

It was determined by measurement that the original efficiency was 29% driving a 50 ohm load and 23% driving a 29 ohm load. After modification the efficiencies were 73% and 53% respectively indicating more than double the efficiency with the modification.

I hope you will find this useful and let me know if it works for you.


Addendum to Ramsey Mods

I originally omitted an important step in the modifications. The PDF has been edited but in case you downloaded the old copy here is the change.

R29 must be shorted to lower the DC bias and final input power adjustment range.

Also, here's more detail on how to measure the DC final input power. Measure the voltage to ground at the test point where R33 and C23 connect. This is VDD. Measure the voltage to ground at the drain of Q3. This is VDS. Calculate the input power as PIN = VDS*(VDD-VDS)/(R33+R34). If R33 and R34 anre unchanged in your unit this is PIN = VDD* (VDD-VSS)/20. This power is adjusted by means of R23.

This equation is in the report in a different form so I thought I would present it here as well.


To Be Tried

Thank you Neil Radio8Z for sharing your research paper on improving the AM25, which I look forward to employing as I bring the Ramsey back up from storage.

Carl Blare

No one is perfect

Thank you for that report. It will be a pleasure to replicate the results. Now can you also provide a report on the antenna setup and its configurations and mounting scheme? That will have an important factor in these numbers.

Obviously unless other's 25 unit is connected to an antenna closely matching your antenna configuration..these numbers wont mean a thing.

In order to replicate the efficiency numbers you have here in the PDF as well as real world results, everything will have to be closely duplicated...well maybe with a little I one is perfect..neither is a modification. :)

Thanks for the info again!


More Information on the AM-25 Modificatons


Yes, in order to duplicate this some standard conditions will need to be set. In the testing I used two resistive dummy loads with the intent that these could be used to confirm the data by others. I will share my antenna information shortly but the best way to proceed is with the dummy loads since duplicating my antenna electrical characteristics exactly would be very unlikely.

I am going to state a caution about what I called "Mod 3" which was replacing the original MOSFET with a 2N6660. Though this FET worked I found after I wrote the report that the gate voltage is running very near the absolute maximum rating of the device. This could have consequences regarding reliability so for now I am not recommending this change. I am going to revise the report with this information. My attempts to control this voltage resulted in a loss of efficiency so this change is not acceptable to me but it may be of interest for others to try to improve it.

I also tried a using a BJT which gave similar results but the device is a special OEM coded TRW transistor so it will not be possible to replicate this unless I find a JEDEC unit which also works. Using the BJT solves the over voltage problem simply so I will pursue this and report what I find.

I have prototyped a gate drive circuit with a 74HCT00 gate which may solve this problem and, if so, I will report this.

Please let me know if you try this and what you find.


Solution Found

I found a BJT which works very well and which is easily obtained. Instead of using a 2N6660 FET use a 2N3904 BJT. It provides better efficiency than the FET and eliminates the over voltage problem.

Here are the data:

For RL = 50 ohms,

VDD = 3.18 V
VDS = 2.32 V
ID = 43.0 mA
PIN = 99.8 mW
Pout = 73.0 mW
Eff = 73.1%

For RL = 29 ohms,

VDD = 3.01V
VDS = 1.99V
ID = 51.0 mA
PIN = 101.5 mW
Pout = 54.2 mW
Eff = 53.4%

The efficiency for a 50 ohm load is satisfying but the lower efficiency for a 29 ohm load is comparable to that obtained with the FET. Perhaps another type of BJT would improve this but the 2N3904 is easy to get and works well.

I had reported efficiency over 70 percent by using a digital gate to drive the FET but this is no longer necessary since the BJT works with the existing Ramsey driver circuit. I left Mod 4 in place but it may not be necessary.

Sorry for the misfire on my original recommendation but I wanted to disclose the problem I found. This change solves the problem.


Appreciation Goes Twice

Radio8Z, your ongoing discovery with AM25 is very much appreciated as this raises hope that the Ramsey transmitter can be put in full, high quality service, alongside the other good transmitters.

The most valuable aspect (to me) about the AM25 is the fact that it is, basically, a very intelligent circuit, but has a few short comings which open the door on experimentation and upgrading, the kind of project perfect for a true part 15er.

You are nominated for the Part 15 Hall of Fame, which somebody should start so we have yet another cultural foot in the world.

Carl Blare

??8Z Modifications for high efficiency

It would be very nice if a link to a schematic could be provided to show the original circuit and the changes.

I have noticed that the efficiency of a Part 15 AM transmitter is lower if it is capable of nearly 100% collector/drain modulation than if it is optimized for high efficiency with the carrier alone. What are the modulation characteristics of the original and modified circuit?

Questions About the Ramsey Mod

Ermi, I appreciate your questions and I will try to respond with a schematic and other details. It may take some time but I will put this on my list.

Re the modulation question, this needs to be pursued. My first impression by listening and watching the envelope waveform on the scope is that the modulation level is near 90%. Before the mods I spent some effort to check the modulation characteristics by using a trapezoidal display on a scope which indicated an acceptably linear pattern. It did pinch a bit above 90% so I set the gain to peak at 90% with audio programming for normal operation. This needs to be repeated for the modified circuit and I will do so. I hope to have a good answer for you as time permits.

It is worth mentioning that I have previously posted that the modulation with this transmitter became distorted above about 70%. This was before I changed it to operate at 100 mW after which I discovered that an output coupling capacitor was leaking which caused the low frequency drain voltage signal to be superimposed on the RF giving the appearance of a badly distorted envelope. I replaced the cap and tested again with a dummy load and the signal displayed as expected up to at least 90%. The distortion was probably not there initially and was just a measurement artifact caused by the leaky cap.


AM-25 Modified, Maximum Modulation

In response to Ermi's earlier question the measured undistorted maximum modulation produced by the AM-25 as modified and driving a 29 ohm dummy load is 90%. Above this there is flattening on the negative excursions.

A trapezoidal display showed very linear modulation characteristics up to the 90% limit both with the dummy load and the antenna connected.

I/ve attached scope screen shots of the modulation envelope with the input audio (440 Hz sine) displayed dual channel mode at 90% modulation. The trapezoid pattern was taken with overmodulation so the tail could be seen. The tail begins at 90%. These are attached to the original post in this book.


Beefing up the 25's modulation limits

Excerpts from an article on modifying the AM 25...

Remove resistors R33 and R34 from the modulator output circuit and replace them with a 6-amp 50-volt diode. Point the diode at coil L4. This enhances peak power modulation so audio can be driven louder with less distortion. For the same reason, replace R12 with a 1-amp, 50-volt diode pointing at the base of Q8. (See complete parts list with sources at end of article.)

Install coils L4 and L6 opposite to the instructions. That is, install them in direct contradiction to the kit's instructions for the orientations of the coils set forth in the Ramsey manual at assembly steps 67 and 72 (indeed in utter defiance of the admonishment that says, "it is vitally important that you properly install this part!"). So now the 8-turn end of L4 will go to C18, and the four-turn end of L6 will face R16. This alteration alone will enable the AM-25 to operate at higher power.

Protect L6. Wind it on the binocular ferrite form provided, as instructed, but, prior to winding, put some extra insulation, such as spaghetti tubing, inside the form's double holes. This prevents the wire from shorting against the ferrite, which is likely to occur as, in the process of winding, the wire's insulation gets worn off. Ramsey does not warn you about this problem.

Replace the coil form for L4. This coil will be handling more power now, perhaps 25 watts, and needs to be wound on a larger form with a heavier conductor. The appropriate binocular ferrite form is part no. BN-73-202 from Amidon.

Omit the low-pass filter, that is, L1, L2, and L3 and all of the associated capacitors. This filter circuit only adds complexity, while limiting your ability to tune the rig. It is a feature intended to imitate circuitry in high-power transmitters.

Article can be found here and is the modifications I mentioned before to use the AM 25 as a Carrier Current transmitter with some REAL muster.

I will be rounding up the parts you have suggested for your modifications Neil. As with Ermi's request of a schematic with your modifications that will be very helpful as well. :)

Regarding L4 and L6 in mounting it in reverse, that works real well to help resolve some of the weak modulation limits of the stock 25, but it also dampens the RF in the end without the added modifications as described in the above article, but it WILL bring the stock 25 to a more "acceptable" range for those overly concerned about this thing being a bit too high for Part 15. Ramsey continues to sell this unit with the same components and values as it did 10 years ago. If the stock unit was that much of a problem with its "SLIGHT" over power of 100mW, I am sure the FCC would have forced Ramsey to revise this thing or pull it...which apparently it is not a concern to the FCC or Ramsey 10 years later.

Thanks for the specs and pdf and updates to it. Will round up the parts and wait for your schematic before performing the mods. I may just buy a new kit and during assembly apply your modifications to a fresh unit, see what kind of results we get there.

Again however, if you are using your modified 25 on your real world antenna system..we need the antenna specifics as well because these modifications and efficiency results while feeding dummy loads is rather mute and does not back up a claim of actually improving efficiency of the 25 into a real world outdoor transmitting antenna.

Thanks again! :)


Schematic Posted

I have attached the AM-25 schematic for the modifications to the original post in this book.


Thanks for the interesting link. It may make sense to reverse L6 in this modification since the Q3 base input Z is low and reversing L6 will step this up perhaps to better match the Q5 collector Z. Next time I have it apart I may try this. If you do, let us know how it works.

Hope the schematic helps.


Book Addition..Schematic

The schematic of the original final stage of the AM-25 has been added in the original post here.


Parts gathered...awaiting new 25

Got all the parts collected as well as revised parts. Awaiting arrival of fresh AM-25 unit. Should be this week and if so..will find the time this weekend to put it together with your modifications and loaded with 50 ohm dummy load.

Then I will try it on my 219 setup, which wont be a problem swapping out the C-CUFF board since the 219 setup is built to make swapping exciter boards fast and easy.


Another AM-25 "Trick"


The data I presented in my "report" showed higher efficiency driving a 50 ohm dummy load than obtained driving a 29 ohm load. Since the 29 ohm load is close to that expected with a resonant antenna system I pursued winding a transformer to place between the transmitter output and the load.

I had a ferrite toroid core in my junk box which I used. Unfortunately, it is of unknown type or part number. It is about .4 inches diameter and has a high AL factor which I measured to be 1.5 uH/N^2. It already had 9 turns wound so I used this as the secondary and wound 12 turns as a primary to transform the 29 ohm load up to 50 ohms. The transformer leakage inductance caused a 32 degree phase shift between primary V and I under load (I probably should have used fewer turns) so I placed a 3900 pf cap in series with the primary which brought the angle down to 5 degrees. The transformer insertion loss measured 0.26 dB. Using this the test with the 29 ohm load yielded 65.6 mW into the load with the DC PIN = 100.6 mW. As a check, the 29 ohm load reflected to the primary as 52 ohms with the cap in place. The difference in efficiency driving a 29 ohm load with and without the transformer is 65% vs. 53%.

(A quick calculation for a T440-15 core gave 26 turns on the secondary and 34 turns on the primary. Needs to be double checked but it is based on XL = 4 times the load R of 29 ohms.)

This is only a suggestion in case you wish to try it.


Another AM-25 "Trick"


The Ramsey AM-25 is an “old school” design, probably from a ham radio oriented person. Why they picked that FET for an output transistor is weird and why they put it on a heat sink is doubly weird. Both were probably for “pizazz”. The complexity of the low-pass filter output is very inappropriate. That type of output network was common in old ham transmitters because they were designed to drive a 50 ohm transmission line to a resonate antenna (achievable at higher frequencies).

A MW part 15 transmitter driving a short 3m antenna requires a series-loading coil to resonate with the small antenna capacitance. Resonance results in only the remaining resistive part of the antenna circuit which essentially comprises the loading coil resistance plus the ground resistance. The Q of such a series resonant circuit is high, negating the need for the low-pass filter in the AM-25. You can eliminate all the low-pass filter components C2, C3, L1, C4, L2, C5, L3, C6, and C7, and you won’t see any discernible harmonic difference. Most importantly, you will eliminate losses due to these unnecessary components.

Otherwise, your changes have improved the efficiency of the poor design so you have gotten close to t he theoretical output-transistor efficiency of 75% for a Class C circuit.

A more meaningful measurement of efficiency for a part 15 transmitter is not just the output transistor efficiency, but the overall transmitter efficiency to the antenna. This includes the transistor efficiency and the loading coil efficiency. The loading coil inductance at the high end of the band (say 1.6 MHz) with a typical 3m antenna capacitance of 30pF is about 330uH. This relatively high inductance will result in significantly high coil resistance, and loss of efficiency to the antenna.

So, in summary, what you did to improve the transistor efficiency is excellent. Don’t stop there. Remove the low-pass filter components and concentrate on reducing the loss in the external series resonating inductor.

Old School Design

Phil, your comments are appreciated and you are correct about the filtering action of a resonant antenna system but there was a bit more I considered when I left the filter in place. This is not a rebuttal of what you posted but is offered in case others want to pursue this. One of the features I always liked about the AM-25 design is its ability to work with almost any antenna without the need for antenna or transmitter tuning. Granted, the overall efficiency and thus the radiated power is pretty miserable without a tuned antenna but it did work to cover my yard with an unloaded antenna and I didn't worry about harmonics being radiated.

The output RF waveform was always a sine wave when viewed on my scope from a load of open to almost a short (10 ohms). A FFT simulation of both the modified and unmodified circuits indicates the second and higher harmonics are below -40 dBc with a resistive load. I have listened to the signal radiated from the antenna at the 2nd, 3rd, 4th, and 5th harmonic frequencies with my receiver and hear no signal associated with the transmitter while the fundamental produces 25+dB over S9 on the meter so it appears to be a clean signal. A check with a spectrum analyzer would be nice but none is available to me.

Per your suggestion I may check the efficiency with the output filter removed. I haven't done this since with a dummy load the measured RMS voltage used to calculate the load power would be misleading since it would include harmonic components. If I do this with the tuned antenna I will post the results. I did measure the modified output filter loss as a stand alone circuit to be .2 dB with the 29 ohm load. This compares to a filter loss of over 3 dB with the original filter. The .2 dB loss represents only a few mW loss at these power levels.

The transmitter also serves the function as a high level test generator providing easily measured current and voltage for network experiments so the sine output is desirable.

The circuit is almost operating Class E. I say almost because of the definition of class E which has to do with the filter network and V and I waveforms at the final collector. A simulation of this circuit was very helpful in evaluating the changes and was in good agreement with the measured DC conditions, the collector waveforms, and the efficiency. Both techniques showed that the phase relationship between the collector current and voltage was nearly 180 degrees with some overlap as the transistor turned on. It is this overlap which produces loss preventing 100% efficiency. I tried several things to minimize this and the most beneficial was changing the duty cycle of the base drive signal. I found both experimentally and by simulation that a duty cycle between 35 and 45 % yielded the best efficiency and coincidentally the duty cycle of the AM-25 driver is in this range after the modification.

There is ringing on the collector voltage waveform at just over 3 mHz which results in the V to I overlap but my attempts to eliminate this made matters worse. This would most likely require a different filter functioning as a tank.

Further improvement, including your suggestion, may be possible and time permitting I may pursue this but I am concerned that more tweaking may make replicating the results I obtained too dependent on individual component tolerances and the load and require critical adjustments which I am trying to avoid.

I have measured the performance of this modified transmitter with my antenna and the system bandwidth is 20 kHz. As you have stated, more than half the antenna system power loss is in the loading coil but improving the Q will reduce the bandwidth and possibly cause audio high frequency loss. There are different thoughts I have read about this, one being that the audio response of most newer AM receivers is low, in the range of 3 kHz at 3 dB down in which case a transmitter system bandwidth of 6 kHz would suffice. The Q of my loading coil measured with Ben Tonque's method is 150 so there is room for improvement. I will probably pursue this when I finalize my outdoor antenna and am confident that the modified transmitter will work well with this.

One thing I found to be very helpful in tuning the antenna is to monitor the phase shift between the antenna feedpoint current and voltage rather than tuning for a peak in either. My digital scope displays this angle and setting it to zero is tricky but it is a very sensitive indicator of resonance.


Schematic Error Found and Corrected

An error was found on the schematic for the modified output stage. The values for C3, C4, and C5 were given in nanofarads. The correct units are picofarads. The schematic has been corrected and is marked as Rev 1. The values listed in the report are correct.

Sorry for the oversight.


Making the Mods

This tremendous project posted by radio8Z is being put to work as I upgrade my AM-25 to act as the transmitter for a carrier current project at 970kHz.

As radio8Z says in the documentation, the modifications were made in relation to a test frequency of 1670kHz, so I am opening the question now of what changes I can make to this output network

to make it suitable for 970kHz into 50-ohms.

The LPB coupler I am using is designed for 50-ohm input, and if more RF power is needed I will later build a linear amp with 50-ohms in/50-ohms out.

Comments will be appreciated.

Carl Blare

AM-25 Filter

Here's a link to an online program which I found very useful in designing the "High Efficiency" transmitter and it can be used to calculate new L and C values for modification to the AM-25:

Otherwise, the AM-25 manual lists parts values vs. frequency for the filter.

I determined that a 5 pole filter is sufficiently effective in suppressing harmonics in place of the 7 pole filter used in the AM-25 and it introduces less power loss than the original filter.

If you design a 5 pole filter (two L and 3 C) set the design cutoff frequency 10 to 20 percent above the operating frequency. This will put the filter transfer peak at the operating frequency.

Select a pi filter with Chebyshev response and an impedance from 50 to 100 ohms (not critical).

I also found subsequent to publishing this thread that replacing the original T44-15 cores with T50-2 cores raised the efficiency by about 8% If you change to the T50-2 cores the number of turns for each will need to be calculated based on the new core material. Let me know if you want to do this and I can provide the necessary information on how this is calculated given the inductance.


Type 2 Core

Yes, Neil,

I will build a 5-pole filter using the type 2 core material, T50-2

Carl Blare

Calculating Turns for Toroid Inductor

The key to this calculation is knowing the "A sub L" factor for the toroid material. To avoid confusion with inductance I will just use "A" for this term.

The number of turns N = SQRT(1000L/A). The factor of 1000 is used to adjust the units to microhenries.

The A for type 2 material is 4.9 so for an inductor of 20 uH the turns are N = SQRT(1000x20/4.9) = 63.88 rounding up to 64 turns.

I used #28 enamel insulated wire for the toroids I wound and could get about 50 turns on a size 50 core without overlap. If you find you need more turns than will fit then you can use smaller wire, a bigger core, or change the filter impedance to give a smaller inductor value.


Revision 3 Change toroids

The PDF file in the original post has been revised to include a change in the output filter toroids.


Log in or register to post comments