arossdotme-planb has quit [Ping timeout: 255 seconds]
arossdotme-planb has joined #neo900
arossdotme has quit [Ping timeout: 264 seconds]
dal has joined #neo900
goiken_ has quit [Ping timeout: 252 seconds]
goiken_ has joined #neo900
galiven__ has joined #neo900
galiven_ has quit [Ping timeout: 276 seconds]
dal has quit [Remote host closed the connection]
arossdotme-planb has quit [Ping timeout: 276 seconds]
Airwave has quit [Ping timeout: 244 seconds]
<wpwrak>
DocScrutinizer05: some entertainment for you: page 25 of https://neo900.org/stuff/werner/tmp/ee/pdf/neo900.pdf (updated the LED backlight circuit: new chip, brought back the current monitor, also brought back CABC)
goiken_ has quit [Ping timeout: 276 seconds]
goiken_ has joined #neo900
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined #neo900
Airwave has joined #neo900
goiken_ has quit [Ping timeout: 260 seconds]
goiken_ has joined #neo900
xman has quit [Quit: Leaving.]
_whitelogger has joined #neo900
pagurus has quit [Ping timeout: 276 seconds]
pagurus has joined #neo900
goiken_ has quit [Ping timeout: 240 seconds]
goiken_ has joined #neo900
chainsawbike has quit [Ping timeout: 276 seconds]
goiken_ has quit [Ping timeout: 265 seconds]
goiken_ has joined #neo900
chainsawbike has joined #neo900
liteIRC_ has joined #neo900
ravelo has quit [Disconnected by services]
arcean has joined #neo900
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
ravelo has quit [Quit: liteIRC for Android]
ravelo has joined #neo900
ravelo has joined #neo900
ravelo has quit [Changing host]
<DocScrutinizer05>
wpwrak: I'd add a 2 emitter resistors to the mirror
goiken_ has quit [Ping timeout: 260 seconds]
goiken_ has joined #neo900
<DocScrutinizer05>
the ideal current mirror is .. ideal ;-)
<DocScrutinizer05>
the CABC vs PWM from CPU needs further investigation resp consideration
<DocScrutinizer05>
plus I dunno why but I'd put the low side LED current sense shunt to the current mirror input branch, not to the output
<DocScrutinizer05>
IOW swap display connector pin 28 with 31
<DocScrutinizer05>
oops sorry nevermind
<DocScrutinizer05>
just that mirror is so... pozzling to look at
<DocScrutinizer05>
hFE = 100 for PMBT3906VS, a 10 Ohm should suffice as emitter resistors for the mirror to compensate variation in Vfwd of LED chains (plus 200mV drop on R2503) and pairing of the two transistors in Q?
goiken_ has quit [Ping timeout: 260 seconds]
goiken_ has joined #neo900
<DocScrutinizer05>
(CABC, PWM) is the PWN output open collector?
<DocScrutinizer05>
if yes then it's fine
radekp has joined #neo900
<DocScrutinizer05>
well, actually seems we have no decent specs for the CABC output from LCD controller, so we need all 'option' for the proto to test: at least PU and PD R, and wired.or / wired-and
<DocScrutinizer05>
of course a GPIO configurable PU and PD would be absolutely great, particularly when the PWM output can also be selected to either be open collector or pushpull
<DocScrutinizer05>
what we probably can safely assume is CABC=1 means LCD bright
ecloud_wfh is now known as ecloud
paulk-collins has joined #neo900
<DocScrutinizer05>
what we also probably can assume: when we *augment* (not override) CABC by PWM, then to make the display more dim
<DocScrutinizer05>
this is in opposite to what N900 does
<DocScrutinizer05>
anyway I still see a problem we need to consider with buffer C2502, when we do (generic) PWM of the boost converter. For PWM frequencies >>1kHz the average output voltage will be more in the range set by R2503 than in the range that results in current through the LEDs = 15mA/PWM%
<DocScrutinizer05>
rationale: during PWM-high the bosst converter will try hard (maximum power conversion) to charge C2502 to the level defined by FB threshold and R2503. During PWM-low the LEDs will still shine from C2502 as buffer capacitor
<DocScrutinizer05>
s/bosst/boost/
freemangordon_ has joined #neo900
<DocScrutinizer05>
mitigation: connect low side of C2502 to high side of R2503/FB instead of GND, so FB also senses and limits charge current for C2502 during PWM-high. We need a lower capacitance 1MHz(?) bypass parallel to R2503 then
<DocScrutinizer05>
hmm >>The boost converters run at 600-kHz fixed switching frequency<<
<DocScrutinizer05>
AAAH nevermind :-D >>Alternatively, a pulse width modulation (PWM) signal can be applied to the CTRL pin through which **the duty cycle determines the feedback reference voltage**.<<
<DocScrutinizer05>
so with the boost and its buffer caps all is fine :-)
<DocScrutinizer05>
for PWM (CABC) however we need to consider how to reliably avoid CABC erroneously getting detected as single-wire when it's possibly operating in a range slightly lower 1kHz
<DocScrutinizer05>
the conditions we need to avoid are: low(long), [high(>100µs) + low(>260µs)]<1000µs
<DocScrutinizer05>
the conditions we need to avoid are: low(>2.5ms), [high(>100µs) + low(>260µs)]<1000µs
raoulzecat has joined #neo900
<DocScrutinizer05>
I guess this could get achieved by applying a CPU-PWM for a ~2ms+ during startup that 'spoils' any of the high(>100µs), low(>260µs) conditions, thus PWM-freq >>1kHz
<DocScrutinizer05>
this is probably also the reason behind converter's datasheet specifying 5kHz<f(PWM)<100kHz
<DocScrutinizer05>
for a CABC 100Hz<f(CABC)<5kHz we maybe want to *constantly* apply/mix a CPU-PWM of 0.1% duty cycle and f>5kHz
ecloud has quit [Ping timeout: 250 seconds]
paulk-collins has quit [Quit: Leaving]
ecloud has joined #neo900
<DocScrutinizer05>
>>To use lower PWM dimming, add an external RC network connected to the FB pin as shown in Figure 19.<<
jonsger has joined #neo900
<DocScrutinizer05>
I just wonder if we could connect the two LED strings in series
silviof has quit [Ping timeout: 250 seconds]
<DocScrutinizer05>
TPS61161 can handle Vfwd of >>30V
silviof has joined #neo900
<DocScrutinizer05>
very interesting and fascinating evaluation to determine what happens with a CABC of e.g. 200Hz. ;-)
<DocScrutinizer05>
see 7.3.1 Soft Start-Up
<wpwrak>
(series) yes, we could. i evaluated this and couldn't find a reason why this wouldn't work. however, the boost converter would be some 2-3% less efficient at a lower current.
<DocScrutinizer05>
but we have less efficiency loss on series shunts
<DocScrutinizer05>
:-)
<wpwrak>
you mean the emitter resistors ? i wonder about these. i see that nokia have that, but what do they do ?
<DocScrutinizer05>
the basically increase the amplification of the BJT
<DocScrutinizer05>
thus reducing error from parameter variation
<DocScrutinizer05>
((you mean) also R2503
<wpwrak>
hmm. well, if we need resistors there, then the effect is likely to be in the range of the loss of efficiency. so, go series then and forget about fancy transistor trickery ?
<DocScrutinizer05>
yep
* wpwrak
loves simplicity
<DocScrutinizer05>
plus we have loss only on one R2503
<DocScrutinizer05>
instead of R2503 and "right half " of the current mirror
<wpwrak>
(cabc) alternatively, we could use the 61160A that removed that fancy programming mode and just has PWM. no risk of ending up on the wrong state.
<DocScrutinizer05>
just been pondering a 5 min to suggest exactly than :-))
<wpwrak>
(mirror) i'll just get rid of it :) been staring at all sorts of explanations and variants for too long already
<DocScrutinizer05>
hehe, goo, so it's tps61161A (not 160A) then, and series, and no mirror
<DocScrutinizer05>
good, even
<wpwrak>
there is also a ~160A
<DocScrutinizer05>
but we want do series, no?
<DocScrutinizer05>
so I hoped for an 161A existing
<wpwrak>
either should be fine. the LEDs have Vf = 3.5 V, so 6 x 3.5 V = 21 V, ~60A does up to 25 V, ~1A does up to 37 V (or a little more)
<DocScrutinizer05>
the single wire programming mode is pretty much pintless when we need a CPU PWM *anyway*
<DocScrutinizer05>
pointless*
<DocScrutinizer05>
sorry for bringing it up
<DocScrutinizer05>
>>The TPS61160A/61A enters shutdown mode when the CTRL voltage is logic low for more than 2.5 ms.<<
<DocScrutinizer05>
so f(min) = 400Hz
<DocScrutinizer05>
hmm nope, 1kHz
<DocScrutinizer05>
>> The device can support the frequency range from 1 kHz to 5 kHz, based on the specification, toff . The output ripple needs to be considered in the range of 1 kHz to 5 kHz.<<
<DocScrutinizer05>
still no good
<DocScrutinizer05>
LCD might be as low as 300Hz
<wpwrak>
however, the ~61A means that i get to replace the diode (can't handle the 36 V limit of the booster). which is good, since it's already "not for new designs"
<wpwrak>
oh, and what about ESD protection around the display connector ? nokia religiously put it everywhere. we have none, so far.
<DocScrutinizer05>
definitely we want ESDprot there
<DocScrutinizer05>
Nik pretty much was "meh, that's cruft"
<DocScrutinizer05>
ok, so I guess TPS6116x[A] are all drop in compatible except for dimming protocol and max Vfwd, right?
<wpwrak>
well, he lives close to the southern sea. warm and humid weather. esd isn't much of a concern ;-)
<DocScrutinizer05>
err really?
<DocScrutinizer05>
rather munich, lots of "Foen"
<wpwrak>
(humidity vs. esd) i never even think of it :)
<wpwrak>
hey, he's south from you, and closer to the sea, too :)
<DocScrutinizer05>
which sea?
<DocScrutinizer05>
Starnberger?
<wpwrak>
mediterranean would be the next stop
<DocScrutinizer05>
;-)
<DocScrutinizer05>
there's the alps in between
<wpwrak>
implementation details
<DocScrutinizer05>
hehe
<DocScrutinizer05>
(TPS6116xX) we will use a generic circuit which allows testing alternatives
<wpwrak>
hmm, and i think we'll want a 22 uH inductor now, if we need to allow for much higher voltages. else current goes through the roof
<DocScrutinizer05>
*maybe* even parallel + mirror, if it turns out serial doesn't fly for wahtever reasons
<DocScrutinizer05>
needs just one more 0R to connect CABC on Rfltr(DNP) to CTRL
<wpwrak>
oh, that's there nik's complicated two diode circuit came from
<DocScrutinizer05>
complicated 2-diode? you mean the Zener for OVP?
<wpwrak>
but what's that stuff supposed to do ? why to we need another PWM channel ?
<DocScrutinizer05>
this other PWM channel is capable for frequencies << 5kHz
<DocScrutinizer05>
even <1kHz
<wpwrak>
the A can go as low as 1 kHz
<DocScrutinizer05>
I know
<DocScrutinizer05>
[2016-09-27 Tue 11:21:02] <DocScrutinizer05> >>The TPS61160A/61A enters shutdown mode when the CTRL voltage is logic low for more than 2.5 ms.<<
<DocScrutinizer05>
[2016-09-27 Tue 11:21:17] <DocScrutinizer05> so f(min) = 400Hz
<DocScrutinizer05>
[2016-09-27 Tue 11:23:01] <DocScrutinizer05> >> The device can support the frequency range from 1 kHz to 5 kHz, based on the specification, toff . The output ripple needs to be considered in the range of 1 kHz to 5 kHz.<<
<DocScrutinizer05>
[2016-09-27 Tue 10:42:19] <DocScrutinizer05> >>To use lower PWM dimming, add an external RC network connected to the FB pin as shown in Figure 19.<<
<DocScrutinizer05>
[2016-09-27 Tue 11:23:16] <DocScrutinizer05> still no good
<DocScrutinizer05>
[2016-09-27 Tue 11:23:30] <DocScrutinizer05> LCD might be as low as 300Hz
<x29a>
will you guys be at the ccc congress?
* DocScrutinizer05
prolly not
<wpwrak>
where do the 300 Hz come from ? i don't see them in any of the data sheets
<DocScrutinizer05>
hmm, I've possibly seen them *somewhere*
<DocScrutinizer05>
we simply can't know for sure
<wpwrak>
there's a "300" in the acx565akm DS right above PWM, but that's for thermal resistance :)
<DocScrutinizer05>
no, I didn't do such mistake
<DocScrutinizer05>
some 'press dadashit" had a quote on a wording like "PWM freq 3500 to 500Hz" or whatever, iirc
<DocScrutinizer05>
350*
<DocScrutinizer05>
whatever, we'll keep enough 'breadboard' space on proto_v2 for testing all variants
<DocScrutinizer05>
I still think a 5kHz CPU-PWM intermodulation on top of the CABC will mostly do to avoid any >2.5ms low levels that would suspend&reset the TPS61161
<DocScrutinizer05>
so the true boost-CTRL PWM is a mixed signal of 300(?)Hz high (100% dim) levels and low 0.01%@5kHz-CPUPWM periods
<DocScrutinizer05>
actually realtek 4503 is a complete copy of TI's TPS6116x
<DocScrutinizer05>
down to the wording in datasheet
<wpwrak>
it's a pretty good clone, yes :)
<DocScrutinizer05>
#not *that* good, see ABS MAX Vin
<wpwrak>
with cabc so unpredictable, how about sending it to an OMAP timer input, and synthesizing the PWM in the OMAP ?
<DocScrutinizer05>
that would keep CPU extremely busy for CABC
<wpwrak>
well, still > Vbat(max) :)
<wpwrak>
CABC is some ambient light thing, no ? shouldn't need adjusting all that often
<DocScrutinizer05>
mind you the CABC brightness is dynamically determined by LCD controller based on video content (brightest pixel)
<wpwrak>
oh, okay ...
<DocScrutinizer05>
CAMC = Contrast Adaptive Brightness Control
<DocScrutinizer05>
CABC*
<DocScrutinizer05>
or Content*
<wpwrak>
and ther edoesn't seem to be much of a chance to just source the original tk65604 ... pity, at least we'd know that this one works
<DocScrutinizer05>
it's meant to save power with dim video content where not a single pixel needs full backlight brightness. So when max bright pixel is 50%, why not reduce backlight to 50% and adjust LCD 50->100%
<wpwrak>
meh. get a samsung if you want movies. it also has enhanced explosion and fire special effects :)
<DocScrutinizer05>
hehehehe
<DocScrutinizer05>
ooh, on a sidenote: got a FLIR now
<wpwrak>
now you're armed and _very_ dangerous :)
<DocScrutinizer05>
damn useful device, need to test it with PCBAs in action
<DocScrutinizer05>
it can show 0.1°C temperature diffs
<DocScrutinizer05>
now if only camera sensors for electric fields would exist :-)
<DocScrutinizer05>
we obviously need to fix physics
<wpwrak>
when will they finally make a compact, $10 HDTV neutrino detector ...
<DocScrutinizer05>
also "typical application" p.2 "3*9 LED" W*T*F?!
<wpwrak>
100 Hz (!) - 50 kHz PWM input
<wpwrak>
Vout(max) 27.5 V
<DocScrutinizer05>
sounds good so far, lemme check the DS
<wpwrak>
input voltage as low as 1.8 V. that's a very dead battery :)
<DocScrutinizer05>
as high as 6V ABS MAX :-/
<wpwrak>
7 V !
<DocScrutinizer05>
I *guess* that TPS61042 is pretty much exactly what I originally suggested
<wpwrak>
hehe :)
<DocScrutinizer05>
alas this one has "min PWM duty cycle 1%"
<DocScrutinizer05>
due to log nature of physiological brightness perception, we would want to be able to use lower min brightness. 1% is rather bright still
<DocScrutinizer05>
even 1/256 in N900 is perceived as too bright still, by some users
<wpwrak>
maybe nominally 1/256 is actually > 1% ?
<DocScrutinizer05>
>> CTRL pin accepts a PWM duty cycle from D = 1% to 100%. Duty cycles below 1% are also possible with the restriction that the device is forced into shutdown as the off time of the applied PWM signal exceeds 10 ms.<<
<wpwrak>
whatever that means :)
<wpwrak>
i am a bit curious about the math behind any duty cycle different from 0% and 100% being able to make a signal > 100 Hz stay low for longer than 10 ms, though :)
<DocScrutinizer05>
that measn that with duty cycles <1% the low-time (high-time(min) * duty cycle) drives the device into shutdown state so it comes up from that state with next high pulse of PWM
<wpwrak>
then you'd run into trouble much earlier
<DocScrutinizer05>
Ton Minimum CTRL pulse width to enable 50µs
<DocScrutinizer05>
t(on)
<wpwrak>
yes, but that's shutdown/activate
<DocScrutinizer05>
*100 = 5ms
<wpwrak>
given the toff following it
<wpwrak>
and with fpwm(max) = 50 kHz, and D(min) = 1%, thigh(min) would be 200 ns. very very far from 10 ms.
<DocScrutinizer05>
while t(off) min = 10ms
<DocScrutinizer05>
huh?
<wpwrak>
(toff) and oddly it has a maximum
<DocScrutinizer05>
yeah
<DocScrutinizer05>
well f(pwm_min) is 100Hz, which matches 10ms CTRL=0 for shutdown
<wpwrak>
ha, that's the detection timing. it MAY turn off already after 10 ms, but it MAY take as much as 32 ms
<wpwrak>
see page 9
<wpwrak>
maybe the sentence "forced into shutdown as the off time of the applied" should read "[...] IF the off [...]"
<wpwrak>
then it all would make sense. low duty but high frequency = okay
<DocScrutinizer05>
this may apply to the 'INNER JOIN' of PWM frequency and the shutdwon timing
<DocScrutinizer05>
yes
<DocScrutinizer05>
you can't have low duty with low frequency
<DocScrutinizer05>
or you might drive chip into shutdown
<DocScrutinizer05>
see >>8.3.5 Applying a PWM Signal to the CTRL Pin with an On-Time tp≤2.5μs<<
<DocScrutinizer05>
so yep, this one looks as a viable alternative too
<DocScrutinizer05>
alas not pin compatible anymore :-/
<DocScrutinizer05>
so we go for that one and forget about the rest, just cruft
<DocScrutinizer05>
however see fig.12 to fig.16
<DocScrutinizer05>
then... 3*3mm *cough*
bbr0ther has joined #neo900
<bbr0ther>
:)
<wpwrak>
(3x3) could be worse. and adding a bunch of components to work around limitations of a smaller chip is likely to be even larger
<DocScrutinizer05>
well, that applies only for final, not for proto_v2
<DocScrutinizer05>
we still *might* get away even with the TPS61161DXX
<DocScrutinizer05>
with either a actual CABC freq >5kHz or a CPU-PWM intermodulation
<wpwrak>
let's worry about that when we make more than 100 kunits, and optimizing for cents does get interesting ;-)
<DocScrutinizer05>
TPS6116X has the advantage of very low ripple on LED current
<DocScrutinizer05>
I'm not optimizing for cents, not at all, you should know me by now
<wpwrak>
tps61042, fig. 17 "TPS61042 With Low LED Ripple Current and Higher Accuracy Using a 4.7 µF Output Capacitor"
ravelo has quit [Disconnected by services]
liteIRC_ has joined #neo900
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
<DocScrutinizer05>
yeah "lowER riplle" (than with just 1uF)
<DocScrutinizer05>
still LED is PWM pulsed pretty much 100%
<DocScrutinizer05>
while on TPS61116X it's basically DC
ravelo has quit [Disconnected by services]
liteIRC_ has joined #neo900
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
liteIRC_ has joined #neo900
ravelo has quit [Disconnected by services]
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
qws-user-1229 has joined #neo900
<wpwrak>
i like the 252 mV threshold of the new chip. now we can use a resistor from the E48 scale, no fancy E96 stuff
<DocScrutinizer05>
compare "Functional Block Diagram" of both chips, for the fundamental difference in how they work. The 6116X controls the boost converter directly to smoothly adjust the LED current to what *the chip* integrated from PWM on EN/CTRL, while the 61042 controls the boost only for OVP and LED *constant* current sensed by shunt R via FB, and just does the dimming by direct PWM to the LED current via switching it with a bottom end additional FET
qws-user-1228 has quit [Read error: Connection reset by peer]
<DocScrutinizer05>
now that is really a pretty minor argument (E48 vs E96)
<wpwrak>
E96 is where i begin to feel uncomfortable. even something as simple as a resistor can become hard to source ...
<DocScrutinizer05>
we don't care about such precision here
bbr0ther has left #neo900 [#neo900]
<DocScrutinizer05>
next lower E48 will do perfect job
<DocScrutinizer05>
err next higher
<DocScrutinizer05>
reducing max LED current by a 2 or even 5% is no issue at all, you can't even tell the difference in A-to-B comparison
<DocScrutinizer05>
mind you, physiological brightness perception is log
<DocScrutinizer05>
you hardly notice a difference when you reduce current by 50%
<DocScrutinizer05>
actually that's what is considered the perception threshold
ravelo has quit [Disconnected by services]
liteIRC_ has joined #neo900
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
<wpwrak>
tz tz tz, delivering less than 100.0%. what a waste :)
<DocScrutinizer05>
not for direct A-B comp, but for sure for noticing it as a difference in properties during normal usage
<DocScrutinizer05>
you remember me telling you I "studied" cinema operator
<DocScrutinizer05>
?
<wpwrak>
i wonder if there are PLLs that can just "multiply" a PWM signal. that would make that CABC compatible with more chips.
ravelo has quit [Disconnected by services]
liteIRC_ has joined #neo900
<DocScrutinizer05>
OHNOES!!!
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
<wpwrak>
ah no, you didn't mention that part of your background
<wpwrak>
but in any case, i guess PWM'ing the LEDs directly should still work. i mean, it's not a CRT.
<DocScrutinizer05>
it of course works, but gives us a lot of other issues to consider, like noise introduced to digitizer for example
<wpwrak>
actually, if CABC is synchronized with screen updates, it may even work better
<DocScrutinizer05>
o.O
<wpwrak>
seems hat the TK65604AB6 also PWMs the LEDs directly, not the control loop
<DocScrutinizer05>
could we get a generic testbed circuit/layout in proto_v2 to test TPS61116(0|1)[A] with serial and parallel LED strings, with and without cirrent mirror, in one of the configs of http://www.ti.com/lit/ds/symlink/tps61160.pdf fig19 and fig16 ?
<wpwrak>
sure. maybe add TPS61042 as plan B ?
<DocScrutinizer05>
if you insist, we can get parallel alternative layout for TPS61042 as well, yes
<DocScrutinizer05>
note that per layout, we probably need to duplicate all the inductor and schottky, and make one each DNP
<wpwrak>
i was actually trying to be sarcastic :)
<DocScrutinizer05>
fail
<wpwrak>
indeed :(
ravelo has quit [Disconnected by services]
liteIRC_ has joined #neo900
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
ossguy has joined #neo900
arossdotme-planb has joined #neo900
arossdotme has joined #neo900
arossdotme-planb has quit [Ping timeout: 265 seconds]
<DocScrutinizer51>
the rest is... up to sloppy design
<wpwrak>
ah wait, there's one more R .. above the RC
<DocScrutinizer51>
sorry, mobile. so 4" screen ;)
<DocScrutinizer51>
the gate output clearly to Rfltr with a 0R. inputs 'jumpers'
<wpwrak>
i'll just wait until you can see it. there's a lot of Rs now, and i have no idea what you want in the end
<DocScrutinizer51>
in the 'end' this is sort of an evalboard for LCD backlight
<DocScrutinizer51>
we have no reliable DS for LXD
<DocScrutinizer51>
thus need to plan for all possible cases of CABC
<DocScrutinizer51>
bbl in 5min
<DocScrutinizer51>
then after proto-v2 we pick the best circuit
<wpwrak>
i'd just try to mimick the TK65604AB6 circuit as closely as possible. perhaps with the option of serial LEDs
<DocScrutinizer51>
we want to improve on that
<wpwrak>
(option: the TK65604AB6 only goes up to 18-19.8 V, which isn't quite enough for 3 x 3.5 V. but we'd have considerable headroom with the TPS61042, so serial ought to work)