The AM Forum

I'm making progress on my higher power class E RF deck, and I'm starting to think about the modulator.  Is it feasible to use modified Heising at say 400 watts carrier?

Just for some ballpark numbers, say we're running 400 watts carrier at 46 vdc drain.  Factoring in losses, that's roughly 10 amps.  Modified Heising calculations say to use about a 23mH choke with a 1100uF capacitor. 

After getting the correct "modulation transformer" using a toroid power transformer with the correct step up ratio, will an audio amp drive all this?

I'm preferring this method because it's very simple, passive and not much can go wrong. 

Jon

Considering losses and proper impedance matching, a minimal 350 Watt analog Audio Amp should drive it.

Phil - AC0OB

Jon,

   What you propose will work, just consider a few not too obvious factors:

1.) The transformer primary will be essentially a zero ohm DC short presented to the output of the SS amplifier. Any DC offset at the amplifier output will cause current flow at the rate of E/R.

2.) Any asymmetry in the audio will create a DC offset and cause current flow similar to a DC offset.

One means of addressing these two items is to AC couple the amplifier output. Using something like two big aluminum electrolytic caps in series -+ to +- will work. Some cringe at the concept, and if worried, you can put a diode across each cap to insure each cap isn't ever reverse biased (cathode to +).

3.) If the SS amplifier is stereo, perhaps the outputs can be combined in a bridge configuration. Sometimes there is a button or jumper to do this, and if not a gain of 1 phase inverter can be applied between the driven input and the other input. The advantage of a bridge is a balanced output, and twice the peak voltage available. This would allow you to cut the modulation transformer turns ratio in about half.

Jim
Wd5JKO

with a blocking cap at the input of the modulation amp, there will be no DC appearing on the output so none
on the modulation transformer.

Also that value will be the DCR of the transformer's primary. Decidedly non-zero. Easy to measure

A good question is what is the ratio required for this transformer??


I'm completely unclear on how this would work. There is asymmetry in all sorts of music and voice that is normally
reproduced by SS amps. No DC offsets to be found.



Only amps that are rated for bridged operation ought to be used for this, imo.
You run the risk of exceeding the SOA (safe operating area) of the outputs/supply IF the Z of the load
goes lower than expected.

Also the amp ought to be stable into a reactive load... some are not.

For example a Crown MacroTech will work fine in this application. They're rated for bridged operation, put out a boatload of current, and lots of power - in excess of the requirement for a 400w rig.

Otoh, why not buy one of QIX's PDM modulator boards?




Might be good to ask Steve in the Class E section of this forum?

maybe I missed something but isn't 46V @ 10A a 4.6 Ohm modulated impedance .... just cap couple to the reactor, no impedance changing transformer necessary...

I would want to clamp the output to supply rails with hefty diodes to handle the turn off transients

Good points! So what are your thoughts on making an bipolar capacitor out of two big electrolytics in series? I once used a big Crown M600 to Ultra Modulate a 4D32. I used two big screw terminal electrolytics (something like 10,000uf at 75v) in series -, + to +, -. This seemed to work fine.

A big SS amp running direct coupled off +/- supplies might have a DC offset voltage just like an OP Amp does. If the load is a 8 ohm speaker, there might be 5 ohms or more DC resistance, so a little DC offset is of little consequence. If the load were the low Z primary of a transformer, the DC resistance might be way less than 1 ohm. But as Beefus correctly points out here, we really don't need a transformer in this instance.

Jim
Wd5JKO

thanks, Jim .... the capacitor series connection you mentioned is correct .... I would also add a diode across each capacitor to clamp the reverse voltage swing


          -----|--------| C-----|-----C |----|-------
                 |                     |                |
                 |--------|<------|---->|-----|

anyway, something like this

I admit I am lost on this DC offset thing.

Even with a DCR load of 1 ohm, the DC offset ought to be <100mv. MAX, from
a not so great typical amp.

This is all from output stage imbalance. Not from input current, like one may see in an opamp
circuit.

Assuming a horrendous offset of 1.0v into 1 ohm it looks bad but it's still just 1 watt!

The value of the cap you might use will be a function of the LF rolloff desired WRT the effective impedance
that the "load" (the Class E stage) shows. Probably you neither want nor need as much capacitance
as you might expect. You can build in a low freq rolloff this way, 1st order, so it is very gentle anyhow. Putting that inflection point at ~60Hz puts you down only 6dB at 30Hz. That's not much given that there is nothing much at 30Hz.

Seat of the pants guess is that <1,000ufd maybe as little as 300ufd will be right.

The advantage of this is that you may be able to skip the back-to-back electrolytics
entirely and go with a film cap - of which there are a lot of pretty high voltage/high capacitance caps now
on the surplus market. They will perform much better than the electrolytics. The thing you may need to be wary of is the breakdown voltage, depends on if the caps will ever "see" any RF or not.

Unclear on where you would put a "direct" feed. Would you be running through an RF choke?

Schematic sketch might be nice...

Jon

Having built a number of AM transmitters with modified Heising modulators driven by modern solid state audio amplifiers (including a legal limit transmitter employing a pair of GS35b tubes)... I can add a few comments/observations:

1. The modulation resistance of your Class E transmitter... running 46V drain-to-source voltage and 10A average source-drain current, at carrier ... will be 46V/10A = 4.6 ohms (as Beefus pointed out).

As Beefus pointed out, a typical modern audio amplifier, with a sufficiently high audio output power rating (e.g. 600W or 800W) will have no problem driving 10A of peak audio into a 4.6 ohm load (even in bridge mode)... since such an amplifier is essentially a voltage source whose output power is limited by how much audio current it can deliver before excessive heating of the output transistors occurs.

2. As you suggest, for optimal low-end performance, you want to choose the Heising reactor (inductor) to have a reactance of around 4.6 ohms at the lowest audio frequency you expect to include in the +/- 3dB passband of your modulator. If the that lowest audio frequency is 31Hz, then:

4.6 ohms / (31Hz x 2pi) = 23.6 mH.

Note: this reactor (inductor) will have to handle at least 10ADC without significant saturation. That's going to be the toughest requirement to meet if you decide to use modified Heising modulation (as opposed to one of Steve's... WA1QIX's designs)


3. You should choose the coupling capacitor between the audio amplifier output and the drain(s) to also have a reactance of around 4.6 ohms at the lowest audio frequency you expect to include in the +/- 3dB passband of your modulator. If the that lowest audio frequency is 31Hz, then:
 
1/(4.6 ohms x 31Hz x 2pi) = 0.0011F = 1100uF

Note: This capacitor should be able to handle a peak voltage of at least 2 x 46V. You will need a 1100uF capacitor that can handle roughly 7A of rms audio current without overheating... e.g. the RMS audio current that will flow through it for several seconds... when applying a 1kHz test tone.

4. Note: Although audio frequency current will be flowing in and out of this capacitor... the voltage across the capacitor (nominally 46V) should not change significantly during modulation. A reverse polarity condition will not occur unless the 46VDC supply is turned off.

5. When the system is first turned on, the 1100uF DC blocking capacitor will be discharged... and the entire 46VDC supply will briefly be applied across the output of the audio power amplifier ... until the capacitor charges up. This may be a problem for some amplifiers... causing the protection circuitry to turn off the amplifier.

6. Personally (although not relevant here, because the voltage across the 1100uF capacitor will never be in the wrong polarity if the 46VDC supply is turned on before audio input is applied to the amplifier)... I am not a believer in the efficacy of constructing AC capacitors from combinations of polarized capacitors and diodes.

7. As Bear pointed out, the audio amplifier should have a negligible DC offset at its output (which is balanced out, during the amplifier turn-on process... before it closes the relays that attach the load). When I was using a ferrite toroidal step up transformer between the audio amplifier and the modulated RF stage (in conjunction with my tube-based modified Heising systems), I was concerned that even a small DC in the primary of the transformer might saturate the core (to some extent). To address/test whether that concern as worth worrying about, I placed a 1 ohm resistor between the output of the amplifier and the primary of the step up transformer... to increase the total series resistance from less than 0.5 ohms to more than 1 ohm. Doing this made no difference in the behavior/performance of the modulator... that I could observe/measure...other than requiring the audio amplifier output power to be increased to compensate for the power loss in the added resistor.

Stu

I may have something like that for him. Chokes from a few decanted Motorola 30A 12VDC power supplies.

If you decide to go with the recommended coupling capacitor linkage,

theses guys have them non-polar in 100 VWV range for 470 uF and 100 uF.

Two 470 uF's in parallel with a third 100 uF will yield 1040 uF.

http://www.specap.com/catalog.pdf#page=24 

Phil - AC0OB

I plan to buy a large Hammond reactor.  They sell a 30mH model at like 20 amps.  It's pricey, but I like new big things.  I must be at my midlife crisis age.   :)

I also got a private email about the rolloff value of a 1100uF cap.  And, as Stu pointed out, it's around 31 Hz.  That's quite low.  This ham used around 470uf for a higher, low frequency rolloff.  For my series capacitors, I've been buying hefty non-polarized models that are meant for wind turbines.  

I read a paper that's on the web about a legal limit, class D transmitter that uses Heising.  One problem he had was that when the carrier drops to zero (the transition from tx to rx), the instantaneous resistance of the "modulation" transformer drops to zero ohms.  This could be bad for the commercial audio amp.  He solved this by putting a large 4 ohm resistor in series with the audio out.  As Stu pointed out, this wastes power, but protects the amp.  4 ohms should be enough protection, so I'll probably do that.

Another consideration that someone mentioned is the step up ratio of the "modulation" transformer.  If modulation resistance is 4.6 ohms, I'm guessing that I want a step down transformer.  So, I'd get something like a 115/60 toroidal transformer and connect the audio to the 115 leads.  Correct?

I've attached another picture of the RF deck progress.  I'm ordering the semiconductors tomorrow.

Jon

Jon

Again... A modern audio power amplifier behaves like a voltage source (having a very low source impedance... much less than 4 ohms) whose maximum output current is limited by the dissipation (heating) of its components.

If you use a step down transformer between the audio amplifier and the transmitter, then you will increase the effective load resistance that the amplifier sees, and you will reduce the power (I.e. V x V / R) that the amplifier can deliver to the transmitter by the square of the turns ratio.

Don't use a transformer in this application if the amplifier is rated to be able to drive a 4 ohm load (or less). If the amplifier is rated to be able to drive a 4 ohm load on each output channel, but only an 8 ohm load when operated in "bridged" mode (both outputs in series, with a common input), then using a transformer with a step down ratio of 1.4:1 might be helpful (if the amplifier is not able to deliver the audio power you need unless you operate it in bridged mode). Again, you would only use this step down transformer if you need to operate the amplifier in bridged mode to get the power you need, and if the amplifier is not specified to be able to drive a 4 ohm load in bridged mode.

A common belief is that a larger value coupling capacitor is better. This is not true. The best low frequency behavior will result if the reactance of the coupling capacitor at the lowest audio frequency is equal to the reactance of the Heising reactor at that same audio frequency. Also, using a larger capacitor will lower the LC resonant frequency, and increase the Q of the RLC circuit (where R= 4.6 ohms) at the resonant frequency to a value higher than 1.... possibly leading to instability (oscillations).

Stu

Get a good amp.   Use a 1 ohm series resistor.   I2R means a 4 ohm resistor is going to eat a LOT more power than a 1 ohm.   Just make sure your amp is 1 ohm stable.

Ive built a decent PWM.   After doing so,  I cannot figure out why you go this route,  the PWM works up to almost 2kv and will modulate anything from 1 watt to kilowatts.

THAT BEING SAID!

I've worked your home brew rigs,  and they sound outstanding.   I totally see why you stick with what works!

But maybe it's time to look at PDM/PWM.....   One modulator,  all projects!   Lol

Can't wait to hear this rig.   I REALLY need a  W.  I. D. E.  AM filter in the Kenwoody.

--Shane
KD6VXI

Fwiw, I voted for a -3dB point of ~60Hz.
Or 50Hz if you wish...

That's plenty low.
35Hz bass is lower than most decent hifi speakers will actually reach!
Lower than the low E string on a bass guitar!! (42Hz. iirc)

And again ur only down technically 9dB at 30 Hz IF the actual -3dB point is 60Hz.

This permits you to halve the inductor and halve the capacitor.

There's really nothing of use in the human voice that low, unless ur "beat boxing"... doing drums
into the mic...

                        _-_-bear

Thanks for the compliments on the audio Shane. By the way, I've scrapped the three 3-500 linear, so I'm back to PW again with my 50 watt class E rig. I need to get this big class E deck going.

Stu, I've read your response three times and I think I've got it. I don't need a transformer if the modulation impedance is 4 ohms (ish) and the audio amp is capable of delivering 4 ohms.

So, I can just skip the "modulation" transformer part of the modified Heising circuit? I just connect the negative lead of the audio amp output to ground, and the positive to one lead of the coupling capacitor (via a large 1 ohm resistor)?

That would save some money.

As for beat boxing bass... That could be a new trend. Tricked out cars with bass thumpers blasting out 40 meters AM. That's my next project... putting bass thumpers in the minivan.

Jon

With all of the complexity, cost and weight of the proposed system, I'm not sure why you don't want to use something infinitely simpler, lighter and certainly less expensive  :)

At the 50 watt power level, just whip together a simple class A series modulator.  Perfect fidelity (it's DC coupled), no audio transformers or chokes and certainly very inexpensive and small.

Or use PWM, which these days is also fairly simple to get going and of course has the efficiency advantage.  It is a little more complex than class A in some respects, actually simpler than pure analog in other ways.

I would avoid audio iron if at all possible, and with solid state like you're building, it's completely possible, and certainly desirable to do this.

Just some thoughts...

Regards,  Steve

Jon

I agree with Steve's comment.

Separately:

If the RF amplifier is off (not transmitting), the audio amplifier will be looking into a series LC circuit consisting of the Heising reactor and the coupling capacitor.

At the resonant frequency of this series LC circuit (e.g. 31Hz or whatever you choose as the low frequency cutoff of the modulator) the impedance that the amplifier is looking into will be essentially the series resistance of the inductor (very low compared to 4 ohms).

Rather than add series resistance (1 ohm might not make that much difference)... I suggest that you use a relay to disconnect the audio input to the audio power amplifier when not transmitting. You can sequence this to happen just before you turn off the RF transmitter. This is what I do.

If you do use an added series resistor (to raise the impedance at the resonant frequency of the LC(R) circuit)... put it in series with the Heising reactor (not between the audio power amplifier and the RF amplifier). That way, the audio power amplifier won't have to deliver additional power to modulate the RF transmitter. The additional power will come from the DC supply (in the form of a DC voltage drop across the added resistor)

With no audio input to the audio power amplifier, I don't think that the audio power amplifier will care about the load impedance (at 31Hz or otherwise) that it is looking into.

Stu

The cost is pretty high for all this stuff.  The audio amp I'm looking at is about $375 and the big choke is $300 plus shipping.  Then comes all the other little stuff that adds up.  All in all, it'll probably cost me about $1000 for the modulator. 

Maybe I will go PWM.  Efficiency is a big seller too.  I'd like to run everything off of a 120 volt outlet.  Steve, you still sell the PWM boards/parts don't you? 

Before I get too into the modulator though, I'll focus on finishing the RF deck.  The heat sink/semiconductor assembly is going to be pretty labor intensive.  I need to drill and tap 18 4-40 holes and then do all the soldering of all that stuff.  I'm glad I made the heat sink assembly removable from the RF deck... it's going to make everything so much easier. 

Jon

Please go PWM, this is what it's all about, is it not?

Simplicity and efficiency, Class E and PWM.

Perfect harmony!

 :)

Steve, it's a 400watt rig...

The K7DYY Super Senior uses an off line switching power supply that is also the modulator. All the schematics are on Bruce's website. Just tossing this into the mix...Also, other then the DIY desire by Jon, an 80m-40m SS is a pretty desirable thing to have.

Jim
Wd5JKO

I've heard the Super Seniors on the air, and quite honestly, I didn't notice any better fidelity between it and any other conventional boat anchor. 

My 2 cents.

Jon

That's not such a good design.  The worst problem is the fact that the modulator output is not isolated from the AC line.

Here's one that DC isolates the output.  It is simpler to use a power transformer and a standard buck regulator (normal PWM modulator), but this one is interesting.

I use this circuit for my 450 watt transmitter at Rattlesnake Island.

(http://www.classeradio.com/off_line_pwm_output.jpg)

If one had say a 1KVA, 120VAC 1:1 input (Isolation) power transformer, it appears one could dispense with U801, T800, D802 through D805, and the separate grounds?

Great circuit BTW.

Phil - AC0OB

Right!  If you want to use a power transformer, this circuit is more suitable.  This is what most people use for pulse width modulation.

(http://www.classeradio.com/standard_pwm_output.jpg)

By the way, the circuit above shows 5 MOSFETs in the modulator.  For 45V @ 10A, only 3 MOSFETs and damper diodes are required.  The circuit as shown is fully populated, and will modulate up to a 1kw carrier.

Here's the audio amp I'm planning on getting:

http://www.sweetwater.com/store/detail/XLI3500 (http://www.sweetwater.com/store/detail/XLI3500)

Just to be sure, because I don't want to fry the amplifier... I can connect the audio output directly to the Heising capacitor once I power on the drain voltage first and let the capacitor charge, correct?

Jon

hi Jon .... that Crown amp is a switcher so the usual caveats about rf radiation and susceptability apply .... consider the end of xmission positive going spike generated by the modulation reactor and how to deal with it

Good point.  I'll sequence both the input and the output of the amp.  Here's the events:

Event #1 - Antenna relay
Event #2 - Drain voltage
Event #3 - Audio amp output connection
Event #4 - Audio amp input connection

Jon

Jon,

   That Crown looks nice. Since your load is around 4 ohms, does 1 channel of the Crown have enough scrote to fully modulate your class E final? Is there a way to parallel the two channels? The Spec sheet says the output is class AB, so the audio channels are linear. The power supply could very well be a switcher.

   I have an old Crown M600. It weighs around 100 pounds, and is rated at 600W / 8 ohms, 1 channel. If you want it, come and get it. (free, as-is)..It is too heavy for me to move.

The Crown M600 Audio Amplifier is a commercial quality amplifier. Frequency range of DC-45kHz with 600 watts (8 ohms) or DC-40kHz with 1000 watts (4 ohms). 0.05%THD.

Jim
Wd5JKO

Jon

At the input of the Heising capacitor (where the output of the audio amplifier will connect), you should add a 25 ohm resistor to ground (wirewound is okay).

This resistor will consume some of the audio amplifier's audio output power .... i.e. about 100% x [1/25]/[(1/25) + (1/4)] = 13.8% of the amplifier's output power.

However, it is needed to provide a path to ground for the Heising capacitor to charge up through... when the audio amplifier is not connected to the Heising capacitor. Without it, there will be a voltage spike (equal to the DC drain supply voltage) across the audio amplifier's output when the switch between the Heising capacitor and the audio amplifier's output is closed.

Stu

Thanks for the offer Jim, but the output per channel on this amp is over a KW.  It should be able to handle it.

And thanks for the tip Stu about the 25 ohm resistor.  I'm not an electronics genius (I copy a lot), but I was thinking about that voltage that would be felt at the amp's output terminals.  This makes sense.

I should be ordering all the big stuff in a couple weeks, so I'm hoping to have this on the air in about a month. 

Jon

With respect to the rate of change in the current through the Heising reactor when the drain voltage DC supply is disconnected from the bottom of the Heising reactor:

A diode to ground, capable of handling a spike in forward current whose peak value is equal to the average drain current... e.g. a 1N5408... can be placed between the bottom of the Heising reactor (DC voltage input side) and ground.

With the DC supply connected, this diode will be reverse biased. When the DC supply is disconnected from the bottom of the Heising reactor... this diode will become (briefly) forward biased... and it will discharge the energy in magnetic field of the Heising reactor ...without producing an arc across the switch that disconnects the DC supply from the Heising reactor.

Stu

Let me propose something different...

For my particular setup, I was planning on disconnecting the drain voltage after the Heising choke.  This would leave the Heising choke and capacitor energized all the time.  Any changes in modulation would be cut off before the drain voltage disconnects via a sequencer disconnecting the audio amp's input and output.

For this situation, wouldn't I then put the 1N5408 reverse biased on the output of the Heising choke (nearest the drain) to ground?  This would also quench an arc drawn on the relay disconnects.

During initial testing, I was using a standard household 15 amp house switch to disconnect the 48 volts.  It was in a regular blue plastic electrical box.  Every time I disconnected the voltage at 400 watts, there'd be a blue sizzle-lean like arc being drawn in the box.  Although, my hand making the disconnect is a lot slower than a relay.  

I ordered a hefty relay for this (regular, not a vacuum one).  

Jon

Jon

The approach you describe would not prevent a discontinuity in the current through the Heising reactor. The diode would always be reverse biased (never conducting). Opening the switch between the Heising reactor/Heising capacitor connection point and the FET's drain (assuming the output of the audio amplifier has been disconnected) would cause a large voltage surge between that connection point and ground. I.e. a 10A x 25 ohm peak voltage surge across the 25 ohm resistor that is located on the audio amplifier side of the Heising capacitor and ground (assuming that the 25 ohm resistor is present). If the 25 ohm resistor is not present, opening the switch would produce destructive reverse breakdown of the reverse-biased diode and/or an arc across the switch.

Stu

Ok, will do.

I'm glad we had this discussion because I just ran a pair of 10 guage DC cables from the closet to the transmitter. One for the drain supply and the other for the IXDD bus. I was just about to cut the drain supply line near the desk to insert the relay. Now, I'll incorporate it into the rack assembly in the closet and leave the cable runs untouched.

Jon