 |
| Original FRG-7700 power supply board. |
I fully expected to find an improved power supply design somewhere, and was little surprised when I didn't. Not that it's an especially difficult thing to design, but simply because these sets are still quite popular (check out the ridiculous prices on eBay!), are now getting on into respectable middle-age, and many would be owned by people unfamiliar with electronics. The only example I could find is Wim's design in his Survival Guide - and it's deliberately more of an update/copy of the original design, with the 13.5V rail still unregulated but a LM317 regulator IC replacing the zener/Q02/Q03 discrete regulator of the original. I'm not sure why there's no other designs for improved supplies - maybe it's as simple as most sets are in the USA with 110/120VAC mains or, like Wim, are in Europe with 220V mains? Perhaps the transformer's primary voltage select taps are better suited to those voltages rather than Australia's 240VAC, resulting in the 13.5V rail being somewhere near what it should be?
Whatever the reason, as I said in my last post I wasn't happy with the 13.5v rails being unregulated and running high, and that a new power supply based on a couple of 3-terminal regulators should be simple enough. Little did I know…
 |
| "Improved" power supply? |
On the bench, it worked beautifully - the BU13.5v rail measured at 13.4v, and the switched 13.5v & 11v rails were soon adjusted to be spot-on. So I installed it in the set, plugged everything in, and turned it on.
The set lives! The front dial lit up, all the controls worked, I could tune in the local AM stations, so I did a quick check of the voltages again. The 11V was spot on - but the two 13.5V rails were down around 12.6V. I measured the voltage across the rectifier - 15.37V.
Damn! I'd forgotten about regulator headroom…
Which is a dumb, beginner's mistake. The fact is that normal linear regulators require volts in to be a certain amount higher than volts out, simply in order to regulate. This is usually called the headroom or dropout voltage, and is often listed in datasheets as VDROP or VDO. For a fixed-voltage regulator like the LM78xx family, VDO is typically 2.0V; for LM317 adjustable regulators, VDO is typically 1.2~2.7V depending on temperature, current, input voltage, etc. And those figures are for genuine name-brand (e.g. TI, LT, Fairchild, etc) devices - who knows what you're getting from your local fart-toy-shop, eBay, DX, etc, whether they're genuine parts or fakes, and what their actual specs are?
(C'mon - did you really think you were getting genuine name-brand quality LM317's for $2.20 / pack of 10, with free shipping?)
And what happens to the unregulated input from the transformer when you put a load on it? It sags, that's what. In this case, it sagged low enough that the 13.5v regulators dropped out of regulation. Time to look at things properly…
So, even though the 'regulated' 13.5V rails of my first PSU attempt weren't actually regulated, to get an idea of the input voltage I had to play with I went back and measured the voltage across the rectifier and main filter capacitor under various conditions:
- No load (J03 & J04 disconnected) : 17.95V
- Standby (J03 & J04 connected, radio turned off) : 17.65V
- Operating (radio turned on, display at full brightness, normal volume, etc) : 15.35v
Anyone know what the load currents are for the various supply rails in an FRG-7700?
The collected wisdom of the Internet is strangely silent on this, beyond vague guesses of the overall current based on the user & service manuals (e.g. "AC 33VA" or "A supply capable of providing 13.5V DC at 1.2 amp (min) is required"). So I decided to measure the current for each rail under various conditions:
| Status | BU13.5V (J04) | 11V switched (J04) | 13.5V switched (J04) | 11V switched (J03) | |
|---|---|---|---|---|---|
| Standby (display off) | 55.2mA | 0mA | 0mA | 0mA | |
| Standby (display on*) | 85.8mA | 0mA | 0mA | 0mA | |
| On (dim, normal volume†) | 67.8mA | 454mA | 12.32mA | 330mA | |
| On (bright, normal volume†) | 73.5mA | 480mA | 12.32mA | 330mA | |
| On (bright, 1/2 volume) | 73.5mA | 480mA | 70mA | 330mA |
† 'Normal' listening volume for me is somewhere around 2~3 ticks on the scale.
That's for the various AM modes; current didn't change noticeably between narrow/CW, medium, wide, or USB/LSB. Switching to FM adds another 20mA on J04's 11V rail, but that's the only difference. My set doesn't have the optional memory module, so I've no idea how much extra that draws (though I wouldn't be surprised if it was at least another 100-200mA, mostly on the 11V rail but with a few mA on the BU13.5V rail).
So, worst-case values are ~150mA total for the 13.5v rails, and ~830mA total for the 11v. The 11v isn't a problem; with 15v available I can use a bog-standard LM317. The 13.5v, though, needs a bit more consideration. You can get so-called "Low Drop Out" (LDO) regulators, but "low" is relative. For example, the LM1085 (a common low-dropout regulator available in both fixed and adjustable variants) has a typical VDO of 1.5V - which might be low enough for this purpose if the input voltage is 15V (i.e. 13.5V + 1.5V) or more, but leaves almost no wiggle-room for variations in regulator behaviour, mains voltage, etc. Ideally, I need something better than that.
Since the 13.5v rails only need to supply ~150mA, I went for a TL1963A from TI (also available from Linear Technology as the LT1963A). These are really nice regulators - for one, their ripple rejection is useful up into the megahertz range, which makes them ideal as post-regulators in switchmode power supplies - although defintely overkill for this purpose. The only real problem with them is that they're only available in SOT-223 (small surface mount) & DDPAK (basically, surface mount TO-220) packages, both of which are difficult to heatsink without dedicating large areas of copper to the task. A quick thermal calculation suggested that the SOT-223 version - all I had to hand - would be adequate, though ideally it should have a bit of heatsinking. So I set about designing a version 2 power supply…
Since its basically built from parts in my junk/spares/samples boxes, I can hardly claim it's an optimal design. It's also hardly original; it's pretty much the reference designs from the LM317 & TL1963A datasheets stuck together. The only slightly odd bit is the relay - since the TL1963A supplies both the always-on BU13.5v rail and the switched 13.5v rail, I needed to switch the latter at the same time as the 11v rail from the LM317. To do this I used a 6v relay I had lying around in series with a 180Ω resistor (R7 - yes I know it's 100Ω on the circuit), which is switched along with the LM317 by the front panel power switch.
Oh, and in case you're wondering why the 120Ω resistor from OUT to ADJ on the LM317 isn't 240Ω as the reference designs in the datasheets indicate - well, almost all the datasheets are wrong. Use 120Ω instead. Always. No exceptions, unless you know the correct reason for its existence (hint: it's not simply part of a voltage divider for the Adj pin) - and if you do you're more than capable of calculating its correct value, which is 120Ω...
In the interests of minimising heat-related problems, I decided to use a small stick-on heatsink for the TL1963A and heatsink both the bridge rectifier and LM317 to the chassis . Although it's something of a no-no (due to thermal expansion potentially leading to cracked solder joints or even device packages) I hung them under the board so I could bolt†† them to the chassis using existing threaded holes. I did form the leads into loops to give some thermal / mechanical relief…
Once that was done I connected the transformer leads & front panel power switch connector, & adjusted the voltages. Then I connected the rest of the set (J03 & J04), and checked the voltages again:
Yes, that's Fluke's notorious and nowdays rarely-seen red headed stepchild - the Fluke 19 multimeter! Chinese build quality at Fluke prices!
(That's a little unfair, seeing as internally it's basically a slightly-less-accurate version of the original Fluke 77/87. It's a little less unfair about the build quality though - although it's better built than most, and even approaches the quality of other Flukes, the irreplaceable switch contacts are a weak point and failure a common mode de décès. Mine is the only one I know of in my circle which is still working...)
After running for a couple of hours, I re-checked the voltages - again, spot-on - and measured a few temperatures. The hottest spot was the transformer (55°C), followed by the LM317 (46°C) and then the bridge (44°C). Since the 13.5V regulator was underneath the board and covered by a heatsink I couldn't measure it directly - but the heatsink, adjacent copper area, and top surface of the board above the TL1963 all read between 40°C and 44°C. The main filter cap - identified as a potential problem by others - was around 48°C~50°C. The major contributor to that would seem to be radiant heat from the transformer & chassis, so apart from using a high-temperature rated capacitor there's probably not much to be done about that.
But it lives!
Totally pointless aside: I stumbled across a few old and current eBay auctions for "Yaesu FRG-7700 Rectifier Units", aka PSU boards. According to the sellers at least they're a "rare part", and apparently worth somewhere between AU$30 and AU$60. If you want the whole lot - fuseholder, voltage selector, transformer, board, and switch - that's AU$100+.
Ahhhhhh … OK, whatever. As far as I'm concerned, and as much as I love it, the whole radio is worth not much more than that…
(I'm not stupid though - if I ever sell it, it's going to be for whatever the current market rate is!)
(†† Look here for some informative links and a fairly light-hearted discussion about the difference between screws & bolts...)
No comments:
Post a Comment