开云体育

Hi,

I just replaced SMPS boards in my QMX with a linear regulator board, and
removed the power switch (I always found it annoying). Some tips, if
someone wants to do the same:

1. 3V3 regulator (I used the 78M33 which was normally used to "bootstrap"
the CPU before SMPSes took over) will get bloody hot. QMX draws around 100
mA on this line, which isn't much, but at 12 V it translates to 870 mW of
power dissipated in the regulator. Make sure you provide as much ground
plane for heatsinking as possible. I did, and my regulator is still almost
90°C. I'm not sure how it will behave long term.

2. CPU needs two fake signals: ADC_5V (2.5 V) and ADC_3V3 (1.65 V). The
reason is that it still tries to drive nonexistent SMPSes, expecting
feedback on these two lines. It won't boot without them. Just provide
resistive dividers, like in the original design (resistor values are not
critical, I used 18k).

3. Lines PWM_5V, PWM_3V3 and LIN_REG_EN can be ignored (unconnected).

4. 100 uF capacitors (instead of 470 uF ones used in original boards) seem
to be sufficient. Maybe even smaller ones would be OK.

When it comes to the power switch removal:

1. V_IN can be hooked directly to +12V (but you lose reverse polarity
protection this way).

2. PWR_HOLD signal can be ignored.

3. PWR_ON signal can be ignored, but the original SMPS board has a 100k
pullup (to V_IN) on this line. Make sure your board contains this pullup
as well (although I used a much lower value, 4k7). If you keep this line
floating, then the (very sensitive) gate of Q202 will get triggered by
literally anything.

4. QMX will power on automatically. When you want to turn it off by long
pressing the knob, it will show "Shutdown!" and hang. At this stage, the
MCU releases PWR_HOLD line, but without the power switch this does
nothing. Only power cycle seems to be able to recover from this state
(understandable).

5. CPU seems to save current state (mode, frequency) to EEPROM only upon
shutdown, so if you just remove the power without shutting QMX down with
the knob press, current state won't be retained. This can be a problem or
an advantage.

Hope it helps.

Jerry Gaffke via groups.io <jgaffke@...> wrote:

Signals ADC_5V and ADC_3V3 back to the host should be a bit high by maybe a tenth of a volt.
If the processor doesn't see a high enough voltage,? it waits until it does before running any further code.
It's ok if those signals are a bit high, but there is no margin on the low side.

Thanks! It works now, but if my QMX ever stops booting, this will be what
I'll check first.

I'm amazed the 78M33 works at all, that's a lot of power to dissipate.

It's very hot, but didn't bang yet. I'm testing it with the case open now.
I'll see what happens when I close it.

I'd recommend either going with a bigger regulator (and a heat sink), or
feeding the 78M33 from your 5v linear..

78M33 and 78M05 are on the same heatsink (ground plane), so I guess it
won't matter which one dissipates the heat, as long as the heatsink can
absorb it...

I'd love to use a bigger regulator, but there's almost no space there. Or
is there? Did you fit a bigger regulator?

Chris <chris.rowland@...> wrote:

I used generic TO220 style linear regulators for my linear SMPS
replacement, had to trim the tabs a bit but they are rated for 1A and
while they get warm seem OK.

Do you have a photo?

Jerry Gaffke via groups.io <jgaffke@...> wrote:

Can't imagine how that dinky 78M33 manages to dissipate so much heat
through one or two skinny Sot23 ground pins.

Barely :) But it's only a matter of heat, not of current, so I guess it
should be OK. It transfers the heat to the ground plane quite efficiently.

Here's how it looks like. Tons of solder to increase thermal capacity.
It's uneven, because there are "vias" (from wire) going to the other side
(and other side has a ground plane too, just to dissipate the heat).



BTW, capacitors are a little too large, the display doesn't fit (I guess
I wouldn't be able to close the case without squeezing them). I'll have
to cut the board below them, or order smaller ones. But first the buzzing
problem (described in another thread)...

Now I'm wondering how hot this IC gets when QMX (with original boards) is
left in the bootloader mode, with linear regulator enabled for extended
time.

Chris <chris.rowland@...> wrote:

Do you have a photo?

Try /g/QRPLabs/message/135796 and the other posts in that thread.

Thanks. I see I'm not the only one with concerns...

Interesting that you say that the regulators get "a bit warm" (which I
understand: safe to touch, won't burn you).

I assume there's no magic in using a larger (TO-220) chip vs DPAK with a
large ground plane. The chip inside is similar, the power dissipated is
the same, the heatsinking area is larger with a ground plane than with
TO-220 with a chopped tab. So my ground plane should be actually colder
than your chip, yet it gets to around 80°C, and DPAK (measured at the tab)
to around 90°C...

Of course not immediately. But slowly, steadily, over the course of maybe
minutes, it gets there and settles.

What current draw do you have on your 3V3 rail? Mine is around 100 mA.

BTW, I ran my QMX with these power supplies for 1.5 hour (without a case).
Nothing exploded. So I guess the regulator can handle it. I'm worried
about capacitors -- how long till they fail in this temperature (they're
directly on the ground plane, touching it, actually being pressed to it by
the display)...

I see you're using tantalum ones, maybe I should switch to that as well.

Chris <chris.rowland@...> wrote:

What current draw do you have on your 3V3 rail? Mine is around 100 mA.

I guess mine is about the same.

Unless mine is damaged and excessive...

You seem quite concerned with reliability, are you planning to operate
24/7 in an inaccessible place?

No, I'm always near the radio. I might leave it by leaving the room, but
that's it. I just want it to work, and not worry about it going bang
mid-QSO.

I routinely turn the power down, for bench testing 7 to 8 V is enough.

Testing is one thing, but when TXing I use 12 V.