dslreports logo
 
    All Forums Hot Topics Gallery
spc
uniqs
84
mlord
join:2006-11-05
Kanata, ON

1 edit

mlord to Teddy Boom

Member

to Teddy Boom

Re: Upgrading Firmware on Thompson DCM475

I really like these EZ-Hook XKM grabbers. They latch onto SMD pins quite solidly and without shorting on neighbouring pins or clips.

Expensive, though. And very difficult to solder wires onto. Not an issue for the Logic Analyzer, which has the proper mating connectors, but for this rig.. somewhat fragile.

I've got a shipment of very similar looking clips on the way from China, with proper solder tabs. I plan to switch this rig over to those once they arrive, assuming they can latch onto things as well as the XKM grabbers do.

EDIT: there is a variant on the genuine XKM hooks that might work better: XK25SS hooks are designed for easier attachment of patch cables, and look like they could be soldered onto as well.

Cheers

Lothario
join:2009-09-30
Ottawa, ON

Lothario

Member

Thanks Mlord,
Everything working fine so far. I think Teksavvy owes you a free month of service for what you are doing for some of their customers.
mlord
join:2006-11-05
Kanata, ON

mlord

Member

said by Lothario:

Thanks Mlord,
Everything working fine so far. I think Teksavvy owes you a free month of service for what you are doing for some of their customers.

A gesture from TSI would be nice, but they definitely cannot afford my usual hourly rates. My wish is that the pain endured figuring out this fix can help keep others going on TSI Cable. That's why I do the upgrades almost for free.

As much as I'm among the first to criticize their lack of great customer support these days, there's really no good alternative to TSI.

Cheers
mlord

mlord

Member

I wonder what the cause of the 02.08 problem actually is?

Now that we've cracked "fixing" the DCM475 modems, I'm wondering what the actual problem is with the 02.08 firmware. Sure, some people with it suffer frequent disconnects/reconnects, but.. why? And why not everyone with that firmware?

One idea, is that it's related to the hourly DHCP renewals that Rogers is switching to in several areas. I wonder if perhaps the 02.08 firmware doesn't deal well with DHCP renewals.. not much of an issue for month-long renewals like we used to get, but with hourly renewals perhaps..

Anyone got any evidence one way or the other?

Thanks

dissilusion
@teksavvy.com

dissilusion

Anon

I for one can say that the disconnect issues on my 2.08 firmware were definitely related to the channel bonding upgrades in my area. Before then, the modem was rock solid.

Teddy Boom
k kudos Received
Premium Member
join:2007-01-29
Toronto, ON

2 edits

Teddy Boom to mlord

Premium Member

to mlord
said by mlord:

One idea, is that it's related to the hourly DHCP renewals that Rogers is switching to in several areas. I wonder if perhaps the 02.08 firmware doesn't deal well with DHCP renewals.. not much of an issue for month-long renewals like we used to get, but with hourly renewals perhaps..

The general theory until now is that the hourly DHCP renewals are put in for areas undergoing upgrade to 4 channel upstream bonding, and that Rogers switches the area back to 7 day renewals after the upgrade is complete. You'll see some mention of that in the 4-channel bonding threads.

The problem with the 2.08 firmware? Who knows.. Last year, or more like the summer of 2011, it was believed that the 2.08 firmware was actually causing some disconnect problems. Like, the 2.08 firmware didn't work well with some CMTSs, in particular associated with Brampton. I think this is the main thread:
»Modem Reset Everyday in Brampton + Slow Download Speeds

Here's another:
»[Cable] BRAMPTON Disconnects DCM 475

Edit: TSI Martin should be able to confirm exactly though
Teddy Boom

Teddy Boom to dissilusion

Premium Member

to dissilusion
said by dissilusion :

I for one can say that the disconnect issues on my 2.08 firmware were definitely related to the channel bonding upgrades in my area. Before then, the modem was rock solid.

Which is great to know, but.. It is important to remember that upstream channel bonding is just a lot harder to do than anything that went before. So, the issue you saw effected all docsis 3 modems regardless of firmware.

What did you do to fix your issues? Did a firmware upgrade improve things, or did a technician come out and fix signal levels? Can you describe a bit about how a firmware was implemented and when the improvement was obvious (like, did Teksavvy ship a new modem)?
mlord
join:2006-11-05
Kanata, ON

mlord

Member

We've got documentation in this very thread of a simple firmware update to 02.16 on the same physical modem clearing up this kind of issue.

I'm (much) more interested in what triggers the issue. Increased channel-bonding is a definite clue, that has been cited many times by many users. It's not clear to me if it's the channel bonding itself though, or the lowered DHCP lease times that accompany it in most (all?) areas. Or some other factor perhaps.

???
nbinont
join:2011-03-13

nbinont

Member

said by mlord:

We've got documentation in this very thread of a simple firmware update to 02.16 on the same physical modem clearing up this kind of issue.

I'm (much) more interested in what triggers the issue. Increased channel-bonding is a definite clue, that has been cited many times by many users. It's not clear to me if it's the channel bonding itself though, or the lowered DHCP lease times that accompany it in most (all?) areas. Or some other factor perhaps.

???

I've got DCM475 with STAC.02.08 and 8x4 bonding (+5dBmV)256QAM down/ (45dBmV)16QAM up. 60 min leases. I don't have a disconnect issue here (anymore).
chrisl83
join:2011-06-21
Almonte, ON

chrisl83 to dissilusion

Member

to dissilusion
My modem was perfect before 8/4 bonding. The day rogers changed it over the modem flaked out on me.
mlord
join:2006-11-05
Kanata, ON

mlord

Member

said by chrisl83:

My modem was perfect before 8/4 bonding. The day rogers changed it over the modem flaked out on me.

Mmm.. okay, so some people show a pretty direct connection between 02.08 flaking out on 8/4, whereas others have no issue.

So no direct guaranteed connection there. That really just leaves a couple of possibilities then: (1) signal levels, except many people with issues report excellent signal stats, and (2) it depends upon the brand/model/version of equipment Rogers deploys at the local node (most likely) or at the CMTS (less likely).

I'm not sure if there's a good way to determine the brand/model of gear at the local node -- maybe nmap could figure it out if we had an IP address for it. But I think (?) the first IP we see in a traceroute is the CMTS, not the local node. Or is it?

traceroute teksavvy.com
traceroute to teksavvy.com (206.248.155.70), 30 hops max, 60 byte packets
1 tomato (xx.xx.xx.xx) 0.419 ms --- my local router/firewall

2 10.124.5.1 (10.124.5.1) 7.628 ms --- CMTS ??
3 69.63.255.189 (69.63.255.189) 15.282 ms --- Rogers router ??

4 fallowfield3.cable.teksavvy.com (24.52.255.78) 8.195 ms
5 fallowfield3.cable.teksavvy.com (24.52.255.77) 13.828 ms
6 2120.ae0.agg01.tor.packetflow.ca (69.196.136.77) 14.127 ms
7 206.248.154.117 (206.248.154.117) 25.048 ms
8 206.248.155.70 (206.248.155.70) 24.025 ms

dissilusion
@teksavvy.com

dissilusion to Teddy Boom

Anon

to Teddy Boom
I first waited for about a month to see if it would clear up, it didn't. Multiple daily reboots. As well I also waited to see if TS was going to announce a company wide directive for the issue. As none was forthcoming, and as the modem was purchased and still under warranty through TS, the exchanged it for a modem with the 2.16 firmware.
nbinont
join:2011-03-13

nbinont to mlord

Member

to mlord
said by mlord:

So no direct guaranteed connection there. That really just leaves a couple of possibilities then: (1) signal levels, except many people with issues report excellent signal stats, and (2) it depends upon the brand/model/version of equipment Rogers deploys at the local node (most likely) or at the CMTS (less likely).

Possibly also (3) the upstream modulation used for the upstream bonded channels. I have 16QAM up, but others have 64QAM up. 64QAM is a bit more work and a different code path on the modem.
mlord
join:2006-11-05
Kanata, ON

1 edit

mlord

Member

Re: Upgrading Firmware on Thompson DCM475


E-Z Hook XKM vs. generic
said by mlord:

I really like these EZ-Hook XKM grabbers.
...
I've got a shipment of very similar looking clips on the way from China, with proper solder tabs.

My first pack of "generic clips" arrived today ($4.60 for 20, compared with $3/one of the XKMs). They are shorter, and fatter than the XKM grabbers. Still (just) small enough to use for flashing the modems, but too fat to fit under the heatsink when needed. They do have the advantage of accepting directly soldered wires, rather than requiring a mating female clip as with the XKMs.

So I'll continue with the XKMs on the flashing rig, but these new clips will replace most of the other test clips and alligators on my bench -- I like the dual grabber action compared with the normal single-hook style of mini-clips.

Teddy Boom
k kudos Received
Premium Member
join:2007-01-29
Toronto, ON

Teddy Boom

Premium Member

The heatsink is sometimes placed very very badly, like covering more than half of the flash chip. On the other hand, the heatsink is made of a composite material that doesn't leave all kinds of metal shavings behind when you Dremel it.
mlord
join:2006-11-05
Kanata, ON

mlord

Member

Heh, looking under the heatsink, one can see that it covers the flash chip when centred on the CPU chip, so one can understand how it might get placed that way. But covering the flash chip is a big no-no IMHO. Perhaps a deeper and less broad heatsink could have done as well without preventing access to other components.

Cheers
Expand your moderator at work
mlord

mlord

Member

Re: Upgrading Firmware on Thompson DCM475

Did another modem update tonight (must be close to double digits by now, but I'm not keeping an exact count of them).

This one was strange. The insides of the modem were a little different from the others: Same PCB layout, routing, pads, etc. But no RF can around the tuner area, only one electrolytic capacitor populated, instead of the usual four, and no ferrites on the power input.

So a "cut cost" edition of the insides.
After an initial goof by me (clip on wrong pin of chip), we got the modem updated fine, and then I soldered in another large electrolytic capacitor to keep the original one from getting too lonely.

Forgot to get a pic of it though.
mlord

4 edits

mlord

Member


Macronix flash chip

Cost-reduced version of board
I just finished updating yet another DCM-475 modem here today, and this one was different from the others. Firstly, it is one of the "cost reduced" versions, with some shielding, capacitors and ferrites omitted, presumably to save money.

But it also sports a different brand/model of flash chip: Macronix MX25L3205DMI-12G.
Here's a link to the datasheet: »www.macronix.com/QuickPl ··· 25L3205D

Fortunately, this chip implements a superset of the instruction set of the original chip, so the same erase/program/read logic still works fine on it. The only reason I noticed is that the program code checks the flash chip ID before proceeding, so it saw the different chip and refused to continue until I updated the code.
mlord

mlord

Member


Reinforced USB micro-B jack
I've now seen several more of the "cost reduced" modem versions in here for upgrades. They seem to work as well as the fully populated boards, so I guess it's not a cause for concern.

As far as I know from people who've contacted me afterwards, the firmware updates do seem to cure the "intermittent connection" issues, so I guess it's worthwhile to keep doing them for a while longer.

I've abandoned the "generic clips" for the flashing rig and gone back to 100% E-Z-Hook XKM Grabbers for hooking up the wires -- they're just so much easier to connect, and they don't let go until I want them to.

The Sparkfun "Pro Micro" board is still going strong, though it did need a repair after a couple of upgrades this evening -- the surface mounted USB connector predictably finally came off it's moorings. So I've soldered it back into place, and beefed up the attachment to prevent a recurrence of that particular issue.

Cheers
-ml
mlord

mlord

Member

said by mlord:

Heh, looking under the heatsink, one can see that it covers the flash chip when centred on the CPU chip, so one can understand how it might get placed that way.

One repair I got to do the other day, was replacing the heatsink inside a DCM-475 modem. The original ceramic heatsink was cracked, and as a result probably not cooling as well as it ought to.

After much cracking and prying, the rest of it came off in a hundred little bits, and then the incredible double-stick mounting tape was peeled off in strips, revealing the Broadcom chip under it all.

And.. revealing a smaller silk-screened outline for the heatsink as originally designed: a standard aluminium block with a multitude of fins, similar to what one might see covering the southbridge chip on a PC motherboard, or an older low-end video card. There are even two small mounting holes on the PCB for the heatsink spring clips.

I had some salvaged heatsinks of the right size on hand, but no clips, so the replacement got glued in place with "heatsink glue". Looks much neater and tidier than the ceramic monolith did, and the modem survived a 24-hour smoke test without complaint.

No pixs (forgot).
mlord

mlord

Member

said by mlord:

The final version of the flash rig. Pretty darned simple.

I've now got a simplified flashing rig on the breadboard.
Simpler? How?

Well, this one doesn't require a PC at all. I've coupled a raw Atmel ATMega328 chip (aka. "arduino") with an SD Card socket. Put a firmware.img file onto an SD Card and the "arduino" reads it and flashes the modem with it.

No PC, no slow serial connection. So the flashing now takes under three minutes for a 2MB image. Still glacial compared with the theoretical speeds, but 2X the speed of the earlier setup!

I'll probably solder up a board with this one on it, to free up the breadboard for other uses. Some pix coming later.

-ml
mlord

1 edit

mlord

Member


The latest SD-Card based flashing rig.
said by mlord:

I've now got a simplified flashing rig on the breadboard.

This is the latest incarnation of the flashing rig.
Standalone (no Linux PC required), turnkey.
Requires a DC power source (4V - 16V), and not much else.

The firmware.bin file is read from the SD-Card and written automatically to both flash partitions, if not already in each.
mlord

3 edits

mlord

Member

Click for full size
The latest SD-Card breadboard arrangement.
SD-Card socket, Atmega328p microcontroller, 3.3V PSU, and a couple of LEDs.

yu13096
@108.171.115.x

yu13096

Anon

Is there anyone doing this near markham, Ontario?

TwiztedZero
Nine Zero Burp Nine Six
Premium Member
join:2011-03-31
Toronto, ON

TwiztedZero

Premium Member

You'd have to come into downtown Toronto and find Teddy Boom See Profile

yu130960
join:2005-08-03
Stouffville, ON

yu130960

Member

Thanks I got a message into him. I wish there was a simpler at home method to force the update, but oh well.
mlord
join:2006-11-05
Kanata, ON

mlord

Member

Well, I could try accepting a "gear deposit" and then mailing the flashing kit (pictured above) to folks, then refunding the deposit (less outbound postage and a $5 wear'n tear fee) when it gets returned.

But probably not economical that way. Postage alone would likely be $8+hst each way.

yu130960
join:2005-08-03
Stouffville, ON

yu130960

Member

It would work with a trusted group of technically capable people that didn't mess up your stuff. I would get the gear, pay the deposit and then ship direct to the next user rather than going back and forth (saving the expense).

I don't mind paying a little extra for the convenience.

FYI I am an at home dabbler in electronics having flashed consoles, sat receivers etc over the last 15 years.

Let me know if it makes sense, otherwise I am going to have to make a special trip downtown.

If there are others in Markham, Richmond Hill it may make more sense to pass it around so that mlord makes his money back faster.