HFU HF Underground
Technical Topics => The RF Workbench => Topic started by: Charlie_Dont_Surf on September 12, 2020, 2244 UTC
-
In the process of debugging the myriad of power-on transient issues with my Stretchyman 40 W TX, I managed to kill the Class-D audio amplifier that serves as the modulator (through a transformer). It appeared to develop some instability issues trying to drive the modulation transformer while it was fed through a few feet of test clip wire. I could see it developing a "motor boat engine" sort of oscillation and before I could completely understand the issue and decide exactly how I wanted to remedy it, the output died and it started drawing a lot of DC current. I don't understand exactly why this happened. It is made to drive speakers through several feet/meters of wire but perhaps the impedance of the modulation transformer with the extra L from the test clips and the added parasitic C put it into instability.
I wanted to replace it but since the original supplier to Stretchy does not appear to be selling these any more, I started looking around for replacements.
I don't know if there is one but I am here to tell you one that is NOT a replacement: https://www.parts-express.com/tpa3116d2-2x50w-class-d-stereo-amplifier-board-with-volume-control--320-699 (https://www.parts-express.com/tpa3116d2-2x50w-class-d-stereo-amplifier-board-with-volume-control--320-699)
I found it online. It looks very similar to the original amplifier. I didn't even bother telephoning to see if the dimensions were close because I assumed that the customer rep would not have that information. For $7 US, I just bought it instead. Once it arrived, three problems came up.
- The placement of the volume knob and the audio input jacks are not in the right place to fit the holes in his enclosure. This is pretty much the show-stopper.
- There is no 12 V regulator that you would need to power the DDS and FET driver. Though it does have a switched volume knob and I did find the PCB trace with the switched DC input voltage, so one could connect that to the input of a 12 V regulator.
- Despite my best effort, I was unable to put the module in PBTL (mono) mode. The TPA3116D2 datasheet is clear that the two L inputs need to be DC-grounded when powering on to enable PBTL mode, and then one can take the mono output off the L and R outputs, with the L+ shorted to L- and R+ shorted to R-. In any case, despite finding what I am 99% sure were the two L inputs to the TPA3116D2 and grounding those, I was unable to get the module to work in a mono mode, as described by the datasheet.
Since it didn't fit the enclosure holes anyway, I didn't bother messing with it any further. Since I went to the trouble of making Stretchyman's TX not self-destruct (so far, anyway), I will probably pair it with a different mono amp and put it in another enclosure, since nothing available with sufficient drive seems to fit the existing enclosure.
Just wanted to give people a heads up that this product is not a drop-in replacement.
-
I see to recall Stretchy mentioning some time ago that the audio amp was no longer available, and the only replacement he could find, which could well be this one, doesn't work for the application. It might have actually been in a post here on the HFU?
-
Ya know, a lot of the problems with these amps stem from the use of a transformer in the first place. Why not double the supply voltage, eliminate the transformer and drive the PA direct from one side of the audio amp IC?
I made a thread about this some time back... https://www.hfunderground.com/board/index.php/topic,40717.0.html (https://www.hfunderground.com/board/index.php/topic,40717.0.html)
*edit* I forgot that this is a 40W TX requiring 48V peak mod, and have not yet found an COTS audio amp IC that will run at that supply voltage. Might be time to build a simple PWM modulator?
+-RH
-
For those reading this and trying to understand what is going on, the TPA3116 modules that I have seen seem to generally have their own boost power converter and can work off 12 to 24 V DC, internally converting whatever it receives to 24 V DC to power itself. This means that they produce 24 V maximum on the audio output. Unrelated, but they also usually have reverse polarity protection.
*edit* I forgot that this is a 40W TX requiring 48V peak mod, and have not yet found an COTS audio amp IC that will run at that supply voltage.
That same particular vendor that I linked to has some Sure amps that can do as much as 60 V into low impedances:
https://www.parts-express.com/pedocs/manuals/320-3328--sure-electronics-aa-bk31394-1x1000w-audio-amplifier-board-manual.pdf (https://www.parts-express.com/pedocs/manuals/320-3328--sure-electronics-aa-bk31394-1x1000w-audio-amplifier-board-manual.pdf)
The drive capability is overkill for the Stretchy boxes but hey, it's not too expensive, not physically large and of course, includes protection mechanisms along with the audio LPF, of course.
I considered ditching the transformer and going straight PWM for my "cannibalization" of the Stretchy box but decided to keep the transformer. The cannibalization is already enough disruption for right now (IMO) and I felt like minimizing the amount of messing around. The messing around will come later. :D
Sure appears to make some that can do up to 120 V. (Scroll down on the linked datasheet.)
Might be time to build a simple PWM modulator?
** WARNING: Thread drift. :D **
Oh, I agree that this can be done too. I have some things in the works.
With the number of little Class-D audio chips on the market for the portable consumer electronics applications, there are a number of options. In the portable application arena, they do not use big LPF inductors on the speaker output of the Class-D amp; instead they rely on the speaker and speaker wire inductance to do the filtering, and it's usually good enough that most consumers don't notice or live with the artifacts.
My thought was that since these chips put out a PWM signal but at a voltage and drive current level that isn't going to modulate a transmitter well, simply add a follower/level-shifter after the PWM chip which will do the heavy switching and let that modulate the transmitter thought appropriate LPF filtering. As a simple feasibility check, I obtained an Adafruit PAM8302 module and put that output into the gate of a IRL510 which I had in my supply and chosen because of the low threshold voltage. It worked well enough that I am going to put together a test board with the PAM8302 driving a real FET driver and then a good FET and some filtering to drive a TX B+ input.
I can't be the first person to propose doing this.
-
It has crossed my mind, easy enough to drive a photocoupler/fet driver combination. Then the sky is the limit, power wise. Much cheaper too than a lot of the other solutions out there.
+-RH
-
CH keeps changing the amplifier design for some completely unknown reason. It's now way more complex and not half as useful.
There is a new design than doesn't have the room for a 'M' style 12V reg under the PCB but will need the pot hole filing out (on the case) to account for the new position. Also there's some dumb audio filter on there that gives a huge mid boost, easily remedied by removing it completely! I would have rather sorted out the filter but they wont release schematics, so was only option. The only downside is that the drive for full mod is 1V2 rather than the 0V4 it was with the earlier amp..
For the missing reg you can scratch off the resist, mount the 'M' reg and wire it to the switch.
I've had to buy another 100 and will mod the RF PCB to have the regulator on board (10W version ONLY)
Str.
-
Also there's some dumb audio filter on there that gives a huge mid boost, easily remedied by removing it completely!
Speaking of filtering, boosting and such:
I dropped in a Drok 100 W mono amplifier (Drok model number is on my schematics in case wants to know) in place of the original one I broke and it is not at all happy with your modulation transformer impedance at low frequencies. I have to cut audio frequencies below ~300 Hz otherwise the amplifier freaks out and goes into protection mode at any sort of reasonable output level.
The modulation transformer on its own can pass down to ~130 Hz without too much distortion and not huge attenuation (from what I remember) so I knew that something else was at fault.
To some degree, it may be partially a problem with (core) saturation since the bass cut is less necessary at lower audio drive levels and/or lower RF FET drain current, i.e., lower standing DC current in the transformer secondary.
The other thing is that the RF FET and everything attached to the modulation transformer secondary is inherently a low-impedance at audio so that it can be a low impedance at DC as well. That low impedance on the secondary is even lower as seen at the transformer primary because of the impedance transformation.
So I assumed that the low-frequency impedance seen at the primary, which was probably very low, was causing the audio amp to freak out. I added two 0.75 Ohm, 50 Watt wire-wound resistors in series with the primary and the amplifier appears to be happy because it has at least a 1.5 Ohm load to work with. It's wasteful but you can make up for the gain loss easily and 100% modulation is still achievable. Best of all, the amp no longer freaks out and I don't have to cut the bass as much.
I tried putting some big capacitors across the two resistors to make an R//C network (so that the added R would be "invisible" above a certain frequency) but I was unsuccessful in making any effect. I left the resistors in, without the C.
Seems OK.
-
Yes the MT is too small for low frequencies. I changed the front end caps to roll off the bass under 150 Hz effectively.
I've since modified the opamp filter so it's relatively flat.
I no longer use a MT and use PWM directly.
Str.
-
My thought was that since these chips put out a PWM signal but at a voltage and drive current level that isn't going to modulate a transmitter well, simply add a follower/level-shifter after the PWM chip which will do the heavy switching and let that modulate the transmitter thought appropriate LPF filtering. As a simple feasibility check, I obtained an Adafruit PAM8302 module and put that output into the gate of a IRL510 which I had in my supply and chosen because of the low threshold voltage. It worked well enough that I am going to put together a test board with the PAM8302 driving a real FET driver and then a good FET and some filtering to drive a TX B+ input.
An update on this.
When I first looked into this, I put a follower on the PAM8302 many months ago, I spent most of my brief time looking at it (5 minutes?) more concerned with assuring that the follower was "following" than anything else. I hooked it up again recently to look at some more things and realized that I should have spent more time examining the nature of the smoosh on the oscilloscope screen that the PAM8302 puts out when I looked at it the first time.
It's advertised as Class-D, which I assume is true, but it's for sure not PWM. It looks more like Pulse Code Modulation (PCM) to me. I can't find the USB stick that has the scope image right now but from memory: the output consists of very narrow (im)pulses at something like single-digit microsecond regularity (so hundreds of Kilohertz) and the height of the pulses follows the envelope of the audio input. So it is not a constant amplitude output and not PWM. I assume that they do envelope restoration on the Class-D output (what is probably more commonly known in hobbyist circles as "drain modulation") to get it to follow the envelope of the input signal.
In any case, this changes what I was going to do with the PAM8302; a variable amplitude signal isn't going to get through a FET driver and also isn't going to take advantage of the efficiency of a switching modulator. There may be a time when I decide to make a series modulator with it (PAM8302 + follower + perhaps a level-shifter) but that time is not now so I have set that aside.
Instead, I have a PWM board in progress using the LTC6992 PWM chip and TI UCC5304 isolated FET driver and some other bits. It basically worked the first time I turned it on but there are some issues to work on before I'd be happy using it outside the lab bench.