Some weeks ago, I noted a distinct lack of apparent activity on 2m FM. Now, this wasnt too unusual, as the band is sparsely populated around here anyway save for CB type tools with no respect for any form of operating standards using it as their own private free-band.
But eventually I noticed that although I couldnt hear anything, I could see high signal strength readings! This led me to discover that the squelch had failed!
Thanks to the information from VK4SN near Brisbane, as shown here http://vk4sn.com/ft857.aspx and my trusty old Marconi 2955, ive been able to correct this fault, or at least, relieve the symptoms!
Following the instructions given, I also found like VK4SN that my new firmware values were around 170, far higher than they were before!
For the time being, I will see how it goes in this state. At some point though I suppose I will have to lift the main board (didnt fancy doing that today!) and check Q1063 as indicated. I might wait until I have located a source of this transistor first!
Thursday, 18 October 2018
Tuesday, 9 October 2018
BOINC-A-Phone
This evening ive made the connections to the phone's battery terminals permanent. At the same time, I made up a 2k4 resistor (which is a non-standard value, so made with a pair of 1k2 in series) to mimic the measured resistance between the 2nd contact (NFC antenna) and 3rd contact (ground) on the battery.
With a bit of cutting, it will now be possible to fit the original back cover on again! The 2k4 resistance did indeed turn out to be the devices BSI signal, the pin being shared with the NFC antenna. With the resistor in place and the supply voltage set to 4.35V (the full charge terminal voltage of a 3.8V Li-ion cell) the device is now perfectly convinced that it has a fully charged battery attached.
So my previously defunct and disused Samsung Galaxy Ace 4 is now a fully working BOINC number cruncher! There is however still the issue of heat to deal with. A quick measure with my IR thermometer shows PCB temperature of nearly 50c. Solving this is easy of course - just a fan. But it will take some thinking about to develop a mounting solution.
 |
| Supply and BSI resistor |
With a bit of cutting, it will now be possible to fit the original back cover on again! The 2k4 resistance did indeed turn out to be the devices BSI signal, the pin being shared with the NFC antenna. With the resistor in place and the supply voltage set to 4.35V (the full charge terminal voltage of a 3.8V Li-ion cell) the device is now perfectly convinced that it has a fully charged battery attached.
 |
| All cores running WU's - 100% battery! |
Monday, 8 October 2018
Phenom-enal
Well I wasnt going anywhere today - not after doing the Yorkshire Three Peaks again yesterday! A bit sore to say the least! So, knowing i'd be staying in anyway awaiting deliveries, I thought i'd get the AMD Phenom CPU installed.
Before I even got the anti-static service kit out, a knock at the door put Sams new GPU in my hands. So that was something else to do while I had all the service kit out.
The AMD Phenom X4 Black Edition CPU went in easily enough, although lifting out the old Athlon chip was a bit tricky, resulting in plenty of heatsink compound getting smeared about. I probably put on a little bit too much new compound, but the cooler clamped back down ok.
So with a slightly worried feeling I re-powered the machine - the BIOS saw the new CPU immediately and invited me to enter set-up and check all was well. Windows on the other hand decided that it would be a bit awkward and just be really slow in updating the drivers and restarting, even accusing me of having a non-genuine copy for a while! But eventually it settled down and all seems to be running fine, if a little hot. Idle, the cores can site around 40c, running 100% BOINC WUs it hits 75%, and is on the very edge of thermal throttling (seen it just once while also opening some PDFs). A new Coolermaster Hyper 212LED CPU cooler is on order, so hopefully once thats fitted the temperatures will be a bit more sensible.
I also installed Sams new GPU for him, and did some cable management within his machine. At the same time I put in his spare 1TB SATA harddrive. This led to a bit of confusion as the machine seemed not to detect it! But it could be found in BIOS, and also we discovered in Windows hardware manager, just not in MyComputer? It turned out of course that as it was unformatted it couldnt be shown as it had no assigned drive letter! Soon sorted once we'd located the windows admin tools page!
I decided late that theres no real likelyhood of my ever needing to use my old phone again, so have dismantled the connector strip from the battery. Even in this state, the 2k4 resistance still exists, so this surely must be involved in the battery management in some way. Tomorrow if I get around to it, I will hard wire that resistance into the supply pins, and i'll let it settle for a few hours to see if any effect is evident.
Before I even got the anti-static service kit out, a knock at the door put Sams new GPU in my hands. So that was something else to do while I had all the service kit out.
The AMD Phenom X4 Black Edition CPU went in easily enough, although lifting out the old Athlon chip was a bit tricky, resulting in plenty of heatsink compound getting smeared about. I probably put on a little bit too much new compound, but the cooler clamped back down ok.
So with a slightly worried feeling I re-powered the machine - the BIOS saw the new CPU immediately and invited me to enter set-up and check all was well. Windows on the other hand decided that it would be a bit awkward and just be really slow in updating the drivers and restarting, even accusing me of having a non-genuine copy for a while! But eventually it settled down and all seems to be running fine, if a little hot. Idle, the cores can site around 40c, running 100% BOINC WUs it hits 75%, and is on the very edge of thermal throttling (seen it just once while also opening some PDFs). A new Coolermaster Hyper 212LED CPU cooler is on order, so hopefully once thats fitted the temperatures will be a bit more sensible.
I also installed Sams new GPU for him, and did some cable management within his machine. At the same time I put in his spare 1TB SATA harddrive. This led to a bit of confusion as the machine seemed not to detect it! But it could be found in BIOS, and also we discovered in Windows hardware manager, just not in MyComputer? It turned out of course that as it was unformatted it couldnt be shown as it had no assigned drive letter! Soon sorted once we'd located the windows admin tools page!
I decided late that theres no real likelyhood of my ever needing to use my old phone again, so have dismantled the connector strip from the battery. Even in this state, the 2k4 resistance still exists, so this surely must be involved in the battery management in some way. Tomorrow if I get around to it, I will hard wire that resistance into the supply pins, and i'll let it settle for a few hours to see if any effect is evident.
Saturday, 6 October 2018
Phenom from Kong
Rather surprised to receive a package from Hong Kong this afternoon. Inside of two layers of thick polythene shipping bag, was a bundle of bubble wrap. Inside that, a sturdy cardboard box. Inside that, another bundle of bubble wrap, containing a sandwich of a plastic chip carrier between two rectangles of high density packing foam. Inside that, with a nice square of antistatic poly foam below its multitude of pins, is an AMD Athlon X4 9952 Black Edition 2.5GHz quad-core processor!
I will resist the temptation to tear down the machine right now, and instead wait until after the weekend before installing!
I will resist the temptation to tear down the machine right now, and instead wait until after the weekend before installing!
BOINC crunchaphone power
Well it seems obvious that if you replace a mobile phones battery with the same voltage supply via the battery terminals, that it would power up as if there is a battery present. And, indeed, this is exactly what happens!
Here the phone is being powered from a 12V lab bench supply, via a 99p buck converter module set to 3.7V, connected using a pair of micro test clips to the battery terminals.
As you can see, all four processor cores are running their allocated BOINC WU's. (Ok I admit, it didnt work first time - I had to remember to set the project to run when on battery!)
What is harder to see is the battery level indicator, which shows 17%, and the screensaver view which has the legend "connect your charger"
Ok, so, clearly the phone cannot tell the batteries level of charge. Since the voltage is correct, and slight alterations of the supply up to and including the charging voltage, dont change this, it follows that the other two terminals of the battery - BSI and BTemp, need to be looked at. Its likely that the issue here lies with BSI, which tells the phone what capacity battery it has installed. With no signal, of course the phone cant accurately judge the remaining charge level.
I suspect that this signal is a simple resistor built into the battery, so it remains for me to see if I can establish the value required to fool the phone into thinking it has a 100% charged battery attached.
I can measure 2300ohm between ground and one of the terminals. But nothing measured to the other. Adding a 2k2 resistor between this terminal and ground on the phone didnt change the charge indication, so my suspicion now is that this value is the BTemp thermistor. That should be easy to check - i'll simply warm the battery up and measure it again!
UPDATE - Well, heating the battery made no difference to the resistance. However, I can now explain this. Ive since discovered, and proved to be the case by peeling off the label, that the pin im measuring the resistance to is actually one of the connections (the other being Vbatt) for an NFC antenna hidden under the battery label!
So im still none the wiser how to make the phone think its got a full battery.
Here the phone is being powered from a 12V lab bench supply, via a 99p buck converter module set to 3.7V, connected using a pair of micro test clips to the battery terminals.
As you can see, all four processor cores are running their allocated BOINC WU's. (Ok I admit, it didnt work first time - I had to remember to set the project to run when on battery!)
What is harder to see is the battery level indicator, which shows 17%, and the screensaver view which has the legend "connect your charger"
Ok, so, clearly the phone cannot tell the batteries level of charge. Since the voltage is correct, and slight alterations of the supply up to and including the charging voltage, dont change this, it follows that the other two terminals of the battery - BSI and BTemp, need to be looked at. Its likely that the issue here lies with BSI, which tells the phone what capacity battery it has installed. With no signal, of course the phone cant accurately judge the remaining charge level.
I suspect that this signal is a simple resistor built into the battery, so it remains for me to see if I can establish the value required to fool the phone into thinking it has a 100% charged battery attached.
I can measure 2300ohm between ground and one of the terminals. But nothing measured to the other. Adding a 2k2 resistor between this terminal and ground on the phone didnt change the charge indication, so my suspicion now is that this value is the BTemp thermistor. That should be easy to check - i'll simply warm the battery up and measure it again!
UPDATE - Well, heating the battery made no difference to the resistance. However, I can now explain this. Ive since discovered, and proved to be the case by peeling off the label, that the pin im measuring the resistance to is actually one of the connections (the other being Vbatt) for an NFC antenna hidden under the battery label!
So im still none the wiser how to make the phone think its got a full battery.
From mobile phone to dedicated BOINC cruncher
One of the problems of making an old mobile phone into a dedicated BOINC cruncher, is power. Another is heat. And the third is connectivity.
These can be tackled in reverse order. Connectivity - well, being an old mobile it no longer has a SIM and so no mobile network. But its WIFI and Bluetooth still work. It is of course a very simple matter to just switch on the WIFI, connect to the local network, and start crunching! As the thing is no longer going to be mobile that issue can be put to bed.
OK then, we come to heat. Crunching the numbers for BOINC work units (WU's) at 100% CPU duty, is clearly going to make it hot. Again, solving this is relatively easy - just give it a fan!
The test set-up shown above is being powered by USB and the fan is a 5V USB fan. Its enough to keep the phone and its battery below 40C. USB isnt really the most convenient source for this project, so later i'll be changing to a 12V source, and so to a 12V fan. But for now it does ok. You can see in the photo that all four CPU cores are in use.
A final aspect of the cooling will be to enclose the fan, phone, and any other electronics needed, so the whole project becomes a single small box of tricks.
Which leads us to the main problem - power.
The phone has a 1900mAh Li-ion battery at 3.7V. A few rough calculations based on the USB current readings, show that it would be flat in a little over 3h if used as the sole supply. I also discovered that when crunching, the battery slowly drains even when on USB supply. Only if the battery is allowed to charge to 100% will it then remain so, I suspect due to the charge controller at this point switching the phone over to purely USB supply. But this ive found led to another issue - remove the USB, and the battery level drops alarmingly! That could be that the battery is on its way out of course. Also, even when supposedly charged to 100%, the USB current crept up close to the maximum 1A of the source. Again, probably due to the charge controllers attempt to handle a failing old battery.
So it makes sense then to plan to dispense with the battery entirely. Since I plan to move to a 12V source (as another future aspect of this project is to go green with a solar supply), then I can utilize a low cost adjustable Buck converter to give me a 3.7V 1A supply, and connect this directly to the phones battery contacts. Because I might need a back up phone, I dont want to do this permanently, so will do so using micro test clips. The only issue I envisage with this approach is that part of the BOINC clients control sets the maximum battery temperature for running the projects, and without a battery present there will be no data on this. I may need to either frig a suitable temperature setting using a resistor (a bit of a bodge), or else also add a suitable thermistor to give an actual measurement of the battery compartment temperature. This is preferable as it will be a good relative reading to the cooling system.
These can be tackled in reverse order. Connectivity - well, being an old mobile it no longer has a SIM and so no mobile network. But its WIFI and Bluetooth still work. It is of course a very simple matter to just switch on the WIFI, connect to the local network, and start crunching! As the thing is no longer going to be mobile that issue can be put to bed.
OK then, we come to heat. Crunching the numbers for BOINC work units (WU's) at 100% CPU duty, is clearly going to make it hot. Again, solving this is relatively easy - just give it a fan!
 |
| Test Set-up |
The test set-up shown above is being powered by USB and the fan is a 5V USB fan. Its enough to keep the phone and its battery below 40C. USB isnt really the most convenient source for this project, so later i'll be changing to a 12V source, and so to a 12V fan. But for now it does ok. You can see in the photo that all four CPU cores are in use.
A final aspect of the cooling will be to enclose the fan, phone, and any other electronics needed, so the whole project becomes a single small box of tricks.
Which leads us to the main problem - power.
The phone has a 1900mAh Li-ion battery at 3.7V. A few rough calculations based on the USB current readings, show that it would be flat in a little over 3h if used as the sole supply. I also discovered that when crunching, the battery slowly drains even when on USB supply. Only if the battery is allowed to charge to 100% will it then remain so, I suspect due to the charge controller at this point switching the phone over to purely USB supply. But this ive found led to another issue - remove the USB, and the battery level drops alarmingly! That could be that the battery is on its way out of course. Also, even when supposedly charged to 100%, the USB current crept up close to the maximum 1A of the source. Again, probably due to the charge controllers attempt to handle a failing old battery.
So it makes sense then to plan to dispense with the battery entirely. Since I plan to move to a 12V source (as another future aspect of this project is to go green with a solar supply), then I can utilize a low cost adjustable Buck converter to give me a 3.7V 1A supply, and connect this directly to the phones battery contacts. Because I might need a back up phone, I dont want to do this permanently, so will do so using micro test clips. The only issue I envisage with this approach is that part of the BOINC clients control sets the maximum battery temperature for running the projects, and without a battery present there will be no data on this. I may need to either frig a suitable temperature setting using a resistor (a bit of a bodge), or else also add a suitable thermistor to give an actual measurement of the battery compartment temperature. This is preferable as it will be a good relative reading to the cooling system.
BOINCing On The 'Phone
Recently, a disaster befell me during a PC upgrade, resulting in a tense few days awaiting an RMA decision over £150 worth of SODIMM RAM, a wasted £15 service charge to tell me what I already knew, and the consequent expenditure of close to £600 to build my eldest lad a new PC,
Sams new machine is now almost complete (just awaiting delivery of the graphics card - running on a terrible 128MB 'industrial' card at the moment!) and its AMD Ryzen 5 processor and DDR4 RAM make it a powerful and reasonably future proof machine.
Doing this however showed me just how old this machine that im currently editing this blog from, was getting! But, it works well enough for my purposes, except that I realised I could do with just a bit more power... so, for the sake of a tenner, Ive bulked up the RAM to the max the current CPU can handle (8GB of DDR3), and noticed a quite substantial improvement. But I realised that the motherboard is capable of handling much more processor than whats currently installed, an AMD Athlon 2.7GHz dual core. I now am awaiting delivery from Hong Kong of an AMD Phenom X4 2.5GHz quad core CPU. This should pep the machine up somewhat! Plus, it will mean I could up the RAM even more, as the Phenom can handle more than the Athlon, at least up to the motherboards max.
The failed machine that we replaced with the Ryzen 5 build, was an Acer 'All-in-one', which whilst it was a pretty top end workstation when purchased, was 3 years old, out of warranty, and otherwise really had nothing going for it! When it was first purchased, I had reservations, but kept them to myself! Foremost was - "what happens when it goes wrong/needs upgrading?" well, as we discovered, what happens is you kill it!
On paper, it was clearly upgradeable. In reality, Acer clearly had no intention of letting the customer have any control! We decided to bump the RAM up to 16GB of nice fast matched pair DDR4, which we knew the machine should be able to take.We put up with the rediculous difficulty of dismantling the machine - trying to unfasten the myriad of too short cables from the very heavy screen, removing the millions of screws, heat shields etc, and having to remove the entire main board to access the RAM sockets! (no RAM access panel on the back like a self respecting laptop!). We fitted the new RAM - and the machine promptly died! After much inspection, it was discovered that the spare slot in the RAM socket was deformed (clearly faulty from manufacture) and the SODIMM stick didnt seat correctly - shorting the pins and destroying the main board chipset! All this of course made irrelevant the fact that the numerous rebuild/dismantle cycles needed to investigate the fault led to the key cables being damaged beyond repair!
So the new RAM went back. This was the most anxious part as it had been bought as a gift by Sams grandad, at a very distressing and sad time for the family. I was very relieved when ebuyers techs reported it fine and the refund issued!
So me and Sam spec'd up a new machine, ordered the parts and built it! A little more on that later...
But clearly I doth digress!...
This post is meant to be about phones and BOINC, as per the title! Well, Im coming to it!
As a result of the failed machine, I now found I had a stick of SODIMM RAM spare, so decided to see if I could make use of it in any of my other machines. Now, somehow ive acquired several laptops, all of different ages and capabilities. It made sense to rationalise them. I looked at all the specs (ignoring the Dell Latitude 400 which is my DOS machine!!!) and the RAM, and decided which of the three would be the most capable, swapped RAM about to make it the best of them. This is now my main radio laptop. The next one down, ive yet to do anything with. But the worst of them, a very old Toshiba Equium dual core, still running windows vista, I decided to play with as clearly it was otherwise scrap, and you all should know by now I dont like to chuck useful kit out.
So I have that old Tosh machine crunching numbers as a semi-dedicated BOINC platform. Its still on vista, but I might move it over to a linux build at some point if that will improve the BOINC performance. It was while setting this up, and fixing a few issues with it, I started to wonder about what I could do with my old mobile phone...
My old mobile has been sat on my shack desk for absolutely ages. Its a Samsung Galaxy Ace 4. It was very unlikely to every get used again, and worth nothing to sell. But - it has a 1.2GHz quad core CPU and an Android OS! I had run BOINC on it before, briefly running Quakecatcher and SETI@home.
Recently though my BOINC choice has been medical research. Ive cut back on the SETI work units, and ended the space and encryption projects. Apart from a bit of involvement with Climateprediction (which was the first BOINC project I ever ran) my choices now are Rosetta and Worldcommunitygrid, where I am attached to immune system and cancer research projects. It occurred to me that I should be able to set up the phone as a dedicated cruncher!
So this is what I have done, although it needs a bit of work electronically! I now have both cores of this main machine and its GPU crunching numbers, the old Vista machine crunching, my current phone crunching (on all EIGHT cores!) and my old phone!
Sams new machine is now almost complete (just awaiting delivery of the graphics card - running on a terrible 128MB 'industrial' card at the moment!) and its AMD Ryzen 5 processor and DDR4 RAM make it a powerful and reasonably future proof machine.
Doing this however showed me just how old this machine that im currently editing this blog from, was getting! But, it works well enough for my purposes, except that I realised I could do with just a bit more power... so, for the sake of a tenner, Ive bulked up the RAM to the max the current CPU can handle (8GB of DDR3), and noticed a quite substantial improvement. But I realised that the motherboard is capable of handling much more processor than whats currently installed, an AMD Athlon 2.7GHz dual core. I now am awaiting delivery from Hong Kong of an AMD Phenom X4 2.5GHz quad core CPU. This should pep the machine up somewhat! Plus, it will mean I could up the RAM even more, as the Phenom can handle more than the Athlon, at least up to the motherboards max.
The failed machine that we replaced with the Ryzen 5 build, was an Acer 'All-in-one', which whilst it was a pretty top end workstation when purchased, was 3 years old, out of warranty, and otherwise really had nothing going for it! When it was first purchased, I had reservations, but kept them to myself! Foremost was - "what happens when it goes wrong/needs upgrading?" well, as we discovered, what happens is you kill it!
On paper, it was clearly upgradeable. In reality, Acer clearly had no intention of letting the customer have any control! We decided to bump the RAM up to 16GB of nice fast matched pair DDR4, which we knew the machine should be able to take.We put up with the rediculous difficulty of dismantling the machine - trying to unfasten the myriad of too short cables from the very heavy screen, removing the millions of screws, heat shields etc, and having to remove the entire main board to access the RAM sockets! (no RAM access panel on the back like a self respecting laptop!). We fitted the new RAM - and the machine promptly died! After much inspection, it was discovered that the spare slot in the RAM socket was deformed (clearly faulty from manufacture) and the SODIMM stick didnt seat correctly - shorting the pins and destroying the main board chipset! All this of course made irrelevant the fact that the numerous rebuild/dismantle cycles needed to investigate the fault led to the key cables being damaged beyond repair!
So the new RAM went back. This was the most anxious part as it had been bought as a gift by Sams grandad, at a very distressing and sad time for the family. I was very relieved when ebuyers techs reported it fine and the refund issued!
So me and Sam spec'd up a new machine, ordered the parts and built it! A little more on that later...
But clearly I doth digress!...
This post is meant to be about phones and BOINC, as per the title! Well, Im coming to it!
As a result of the failed machine, I now found I had a stick of SODIMM RAM spare, so decided to see if I could make use of it in any of my other machines. Now, somehow ive acquired several laptops, all of different ages and capabilities. It made sense to rationalise them. I looked at all the specs (ignoring the Dell Latitude 400 which is my DOS machine!!!) and the RAM, and decided which of the three would be the most capable, swapped RAM about to make it the best of them. This is now my main radio laptop. The next one down, ive yet to do anything with. But the worst of them, a very old Toshiba Equium dual core, still running windows vista, I decided to play with as clearly it was otherwise scrap, and you all should know by now I dont like to chuck useful kit out.
So I have that old Tosh machine crunching numbers as a semi-dedicated BOINC platform. Its still on vista, but I might move it over to a linux build at some point if that will improve the BOINC performance. It was while setting this up, and fixing a few issues with it, I started to wonder about what I could do with my old mobile phone...
My old mobile has been sat on my shack desk for absolutely ages. Its a Samsung Galaxy Ace 4. It was very unlikely to every get used again, and worth nothing to sell. But - it has a 1.2GHz quad core CPU and an Android OS! I had run BOINC on it before, briefly running Quakecatcher and SETI@home.
Recently though my BOINC choice has been medical research. Ive cut back on the SETI work units, and ended the space and encryption projects. Apart from a bit of involvement with Climateprediction (which was the first BOINC project I ever ran) my choices now are Rosetta and Worldcommunitygrid, where I am attached to immune system and cancer research projects. It occurred to me that I should be able to set up the phone as a dedicated cruncher!
So this is what I have done, although it needs a bit of work electronically! I now have both cores of this main machine and its GPU crunching numbers, the old Vista machine crunching, my current phone crunching (on all EIGHT cores!) and my old phone!
Possible Future Blog Move
Heads Up,
Due to insurmountable difficulties with Blogger in relation to mobile devices and account tie-ins to ISPs, Im considering my options to migrate my blogs to another blog service.
Those of you who occasionally drop by to see if ive anything new going on, be aware that any links you have may stop working - at least until I tell you what to change them to!
Due to insurmountable difficulties with Blogger in relation to mobile devices and account tie-ins to ISPs, Im considering my options to migrate my blogs to another blog service.
Those of you who occasionally drop by to see if ive anything new going on, be aware that any links you have may stop working - at least until I tell you what to change them to!
Subscribe to:
Posts (Atom)