18LF2550 not working with USB.....

[JAM2014]

Hi All,

I have a 'core design' based on the 18F2550 that I've used successfully on a number of projects that require USB connectivity. Recently, however, a new design utilizing the 18LF2550 (note the 'LF') is not being recognized correctly by Windows..... A review of the 18F2550 datasheet suggests two problems.

1. The problem design uses a Vdd of 3.6V (required by a GSM modem). I'm using the internal Vusb regulator, and I'm measuring 3.3V on the Vusb pin, but at a 3.6V Vdd I'm clearly out-of-spec here. I'm planning to disable the internal regulator, and generate the 3.3V Vusb externally with a zener diode.

2. I'm using a 20MHz crystal, the PLL5 fuse, and the CPUDIV1 fuse, so I'm clocking the PIC at 48MHz. Again, although the chip seems to be working correctly (except for the USB, which is used exclusively for a bootloader in this application), I'm clearly out-of-spec on the maximum clock rate for the chip at this supply voltage. I'm planning to use the CPUDIV6 fuse, and clock the PIC at 16MHz instead.

Is it likely that these changes will fix the USB problems I'm experiencing? I'm traveling now, so I can't test anything for a week or so, but I will have the opportunity to order any needed parts I might need in the interim. I think I'm on the right track with these fixes, but I'd like confirmation from someone that may have faced the same issue previously.

Thanks!

Jack

[PCM programmer]

The "K" version of that chip, 18F25K50, runs at 3.3v right up to 48 MHz.
Check the compatibility with your exisiting design and see if you can
just drop it in.

You will need ICD3 or Pickit3, or the equivalent CCS programmer to
program it.

[Ttelmah]

Have to agree with this suggestion.

The 'K' chip also has far fewer errata than the F2550 versions. Makes it a lot easier to work with!....

Alternatively the J50, will also go into the same socket, and has the advantage of quite a few extra peripherals, that are 'relocatable'. So a second UART for instance.

I'd suspect the Vusb, is spiking down, when the transceiver loads change, hence the recognition problem. It's probably got a characteristic impedance of some kilo-ohms, trying to generate 3.3v, from 3.6v....

Technically, the maximum voltage allowed for Vusb, is 3.6v. So if you can guarantee your supply won't go over this, you can feed Vusb directly from the supply for the rest of the board. However a little extra isolation and smoothing here is well worthwhile.
It is worth pointing out just how high the instantaneous currents can be on USB. When the transceivers switch, for a few nSec, many mA flow. It is this momentary spike that the capacitor on Vusb has to deliver, and 'why' it must have good HF performance.