TeensyMaestro

I just wanted to document my experiences with VFO encoder reliability in my TeensyMaestro, in case its of help to anyone else. The symptoms on my TM were that sometimes on powering up the TM, neither VFO encoder would work straight away. Sometimes they would start working after a minute or two, or I had to power cycle the TM to get them to start working. I also experienced "stuttery" response tuning until the TM had "warmed up" for a few minutes.

My TM is a V2 build using Simon's PCBs. The two VFO encoders are specified with a supply voltage range of 5-24v and when I measured various USB power supplies connected to my TM, the internal volatge presented to the encoders was around 4.8v. This worried me so I removed the 5v pin from the connector of one of my encoders so that I could run that encoder off a decent bench power supply. I measured the operation of the encoder and its current draw against voltage:

• stops working at indicated 4.77v
• works at 4.8v and above
  • 14mA at 4.7v (not working): 66mW
  • 15mA at 4.8v (working): 72mW
  • 16mA at 4.9v: 78mW
  • 17mA at 5v: 85mW
  • 18mA at 5.3v: 95mW
  • 20mA at 5.7v: 114mW
  • 21mA at 6v: 126mW
  • 26mA at 7v and above (current but not power limited): 182mW and up

So this explained my problems - too low a supply voltage for the encoders. I know Len has reported his work down to 3.3v, but I guess it depends on how lucky you get with the manufacturer. So I decided to install a small buck/boost converter inside my TM to power the encoders at a higher voltage. This is safe to do because the encoder data outputs are open collector and those are still run at 5v from the pullup resistors on the PCB. I used this SparkFun PCB: https://www.sparkfun.com/sparkfun-buck-boost-converter.html which has the option to install a leaded resistor to set your output voltage. I used a 470kOhm which sets the output at about 6.4v. I made a flying lead with a 2 pin header and moved the encoder supply pins to a 2 pin socket, you can see the buck/boost converter on the side wall of the case:

Since I have made this modification, my VFO encoders have worked perfectly each time I have powered up the TeensyMaestro, so I've very pleased with how its working now.

I hope that might help if anyone else buys encoders that operate only at their specifications,

Best 73,
Jonathan G4IVV

Hi Len,

The rev 2 hardware has the encoder open collector pullup resistors connected to 5v:

I just pulled mine apart to check, the top end of R1 & R2 is at 4.75v and the IDX_A_PhaseX pins are at about 3.9v high level, I guess being clamped by the Teensy inputs? There's certainly more than 3.3v directly measured on the Teensy 4.1 input pins.

This doesn't look right, I'm thinking about modifying my rev 2 board and moving the 4 pull-up resistors to 3.3v. I wonder what Simon thinks?

73, Jonathan

Thanks Simon,

Is it possible to just use the internal pullups in the Teensy or are they too high an impendance for these encoders? I'm happy to lift the 5v end of these resistors and wire them to the 3.3v supply in the meantime.

73 Jonathan

Hey, great news on the 8600 Len! I do hope you have as much fun as I've had with mine, it's a fantastic rig.

In the meantime I thought I'd see if its possible to do an easy wiring mod in case we need to keep the pull-ups. I managed to do this without having to disassemble everything. I lifted the 5v end of the 4 resistors and found I could pickup 3.3v from the end of R7. Folding the resistors back over the capacitors eliminates the risk of them touching the 5v pads:

Everything is working great, so at least we have that as a fall back mod. Will be interesting to see how you get on with enabling the internal pullups, as then we can simply clip out the 4 resistors.

But before that, get your 8600 up and running 😀

Edited to add: whilst I haven't found an official datasheet, indications from comments on the web suggest the internal pullup when enabled is around 47kOhm. This may mean the decoupling capacitors will be too large (time constant will be then ~5mS). So the capacitors may need to be replaced or reduced. Will be interesting to see if you notice any issues tuning with the 100nF's in place.

73, Jonathan

I thought I'd try and dig a bit deeper to find out why my encoders didn't work at 5v and I had to install a buck/boost converter to get them to start up operate reliably. Whilst searching the web I found this video on YouTube where a chap investigates this problem with exactly the same encoder:

https://www.youtube.com/watch?v=JvUNzHatPu0

So I decided to open up my VFO encoders and sure enough:

Well that 7805 explains why I had to feed in more than 5v to get it to work. In theory it should need about 6.8v to guarantee operation. Rather than simply wire across the regulator as suggested in the YouTube video I decided to remove the 7805's. This is pretty straightforward, the steps I used were:

• cut the two leads
• apply some fresh solder and soldering iron to the heatsink tab
• after a few seconds the solder will flow under the 7805 and it can be lifted off with tweasers
• then desolder the cut lead stubs and clean up

There are 0805 resistor pads conveniently located to bypass the 7805, but I didn't have a suitable low value 0805 to hand. I could have just shorted it out, but as I had some low value leaded resistors (10Ohm) I used those for some in-rush protection:

Final step after reassembling the encoders is to modify the voltage rating on the label with an indelible marker, otherwise I might forget in the future they can only be connected to 5v!

Now both my VFO encoders are working reliably on the original 5v supply and the mystery is solved.

73, Jonathan

I thought I'd take one more go at simplifying the VFO encoder 3.3v Teensy interface fix.

SImon is right that relying on the internal Teensy pullups is unreliable because the time constant of the higher impedance of these internal pullups (we think somewhere in the range 50k-100k ohms) and the 100nF decoupling capacitors is around 5mS, much slower than using the 1k pullups.

I setup my 'scope to measure the risetimes. First, here's with 1k pullups and 100nF capacitors:

Interestingly I noticed that spinning the knob resulted in reduced signal amplitude, because at 600 pulse/rotation even this time constant is too large.

Removing the 1KOhm pullups, but keeping the 100nF capacitors has the expected effect (slower 'scope timebase for this capture):

The rise time is measured at 5mS as expected and not useable.

But what if we reduce the capacitance? I replaced the capacitors with 1nF, which should give reasonable rise time and still provide some RF filtering:

Bingo! 57uS rise time and a nice clean 3.3 volt logic signal from the encoder.

Encouraged by this I replaced all 4 encoder capacitors, so the final modification is really simple:

• remove the pullup resistors
• swap the 4 capacitors for 1nF ceramics

This is now working very reliably for me, and doesn't need any board modifications.

73, Jonathan

Thanks Simon, it will be great to get another verification. Pleased to be able to help and also put my retirement to good use Hi.

73, Jonathan

ps. the more I play with my TeensyMaestro after doing this mod, I think the "fast spin the VFO knob" behaviour is improved with this smaller time constant. Previously I could see a little stutter when rapidly moving up/down the band. Probably limited by interrupt latency and API timings now.

Hi Jonathan,

The PCB for my encoders looks very different than yours - a much simpler circuit. While there is still a 7805, I see mine doesn't have the A7 diodes (D1 and D2) that yours has. Are they connected to the A and B outputs on yours? Perhaps the forward voltage drop of 1.1v is problematic?

Which model do you have, Jonathan? I have a Hakko FR-301, it saves so much time. As long as I clean it regularly it works really well.

Ha! I have the same iron as well. Great news on the 1nF caps. I'll be replacing those later today too. Len's work on interrupt handling has made it very successful. As Neil pointed out, the contributions from the Flex Radio community with this project have been fantastic.