Latest experiments ...

a low voltage DS(hcottky)TL test ::

in :: https://www.falstad.com/circuit/circuitjs.html

in :: https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html

---

j-Fet test ::

---

scs test (a variety modified from the 7-th ed. GE Transistor Manual 1964) ::

---

about :: the following is carried out on intent to find out the low-power-edge function of the scs / pulse-driven logic - - coz there're a lot of old discretes lying around nothing to do with - - in most cases the DTL is out performing such in a sense of power efficiency (but there are few exceptions and some conditional exceptions . . . most yet to be defined . . .)

---

(-a-) -- incomplete design of low power HOLD capable RAM/Counter . . .

(-b-) -- i guess the proper operational voltage levels' range selection allows the following to operate without pulse amp stages . . .

(-c-) -- ↓it↓ draws about 10mA per cascade/stage for about 400kHz CLK src. (& takes an additional circuitry to read the trigger OUTP levels . . . e.g. adding even more power use in real implementation . . . )

---

[Eop]

32kHz X-tal osc -- new variants

12µA Bjt v. ::

2µA j-Fet v. ::

---

see also an old C-MOS inverter variant : https://chpsndtch.blogspot.com/2017/08/cmos-osc-s-test.html

---

[Eop]

a couple of more 5V 32k osc.-s

a tuned (this time) j-Fet amplifier stage

---

[Eop]

yet another 32768Hz Quartz Osc.

Blind tuned to work with some consequences from . . . at 5V v. of it . . .

-- with the single BJT stages and the battery feed and "sensitive"(tuned) oscillators -- it is most unlikely to get anything "properly" to work -- . . . -- moreover with the 1.2 volt battery the grid has to be tuned to work in 3 different areas of operation of the bipolar transistor -- low-voltage (below normal) , normal voltage (for BE CE voltage drops) and hi-voltage (above normal) -- that has to be achieved with fixed biasing elements -- . . . -- in below circuit the critical elements appeared to be C1 , C8 , C4 , C10 -- less C1 more C8 ... also the series connection of C4 and C10 = Cp of the X-tal resonator it first started to "charge" the X-tal with ratio 1.2x : 6x Cp but as i increased C8 the opt. ratio changed to 1.5x : 3x Cp ???

the "above normal" range ↓↓ -- i just fast tuned it just to "charge" the X-tal -- this to see if the current waveform for X-tal goes "trapezoidal" (← failure in a sense of frequency stability -- i assume) or stays Sine . . . . . . . . . well it stayed sine -- but for this one the further frequency tuning and likely some attenuation is a must ↓↓↓↓ (see next) . . .

---

there is a very simple circuit for 8MHz crystal but (it requires less energy for startup) and it also works only with very narrow range of the supply parameters -- so -- it can't be simply adapted for 32k
. . . ?amazingly i've already proven it otherwise ???

8MHz ::

the C14 caused pulsation is likely causing and the instability in the 5v experiment above ↑↑↑↑

32kHz ::

Low NRG 32k ::

Ultra-tuned UHF ::

nearest match from web
http://www.nutsvolts.com/magazine/article/bipolar_transistor_cookbook_part_5

The failed but extremely fast starting UHF resonator Osc. experiment

---

[Eop]

CMOS osc.-s TEST

! notice the word "test" above !!!

in images

originally with j-FET (← to cancel out BJT feedback to quartz) and BJT (← the low power side) while i realized the j-FET can pretty much do the thing on it's own . . .

the copy paste from NXP-s datasheet (the R3 + diodes added as a speculative ¹ startup booster / ² frequency booster / ³ exponential atenuator)

"alternate compile" of the above prev. (not thoroughly studied !!!)

what the NXP d/s does not tell you . . . about 32kHz quartz (the "C5" is the type of magic wand here)
...

Sine wave crystal oscillator concept

i still poorly understand how to frequency tune this thing . . . but the main concept has been achieved -- a pure undistorted sine wave form for crystal . . .

. . . i find it curious that the appropriate web search can't find any analog circuits for such . . . . . . . . . even more curious is the fact that most micro controllers use digital/discreet stages based quartz drivers that for lower frequency X-tal-s as far as i get attempt to over stress the device and eventually ruin or kill it ???
(8 billion people with no brain on this planet ... statistically impossible but is what the prolonged web search stat.-s retrieves - i roger it but i can't get it - it's insane)

---

update 2017.04.10 21:41 UTC+3h : revised an old design -- not as good as previous but seemingly near the sine -- also uses a lot of less power -- it likely would work and with even less power but assumingly with longer startup time

---

[Eop]

Custom eXperimental 0.6 ... 1.6 V inverter

From back to forth --

32k768 Qz Driver ::

the interesting thing about this X-tal pusher is that the 1 single inverter most likely has a sufficient gain and speed to manage the task (if it's already oscillating) but using uneven no. of gates (1,3 were tested) and increasing the +feedback the oscillations start ubruptly at 100x to 300x higher fq. than the one of the quartz -- otherwise the quartz won't get into phase lock with feedback and thus won't collect the energy for start up ???
-- however the shown - an even no. of gates - v. has quite flexible tuning range where everywhere the oscillations do start e.g. the quartz gets phase locked with feedback and starts storing the energy . . . this is the first time i encounter such . . . curious ?

AA(AAA) behavior/capabilities ::

. . . simple things are complex to design ::

[Eop]

another 32768cps osc.

the resistor values can be varied** from R.c ~50k(perhaps less) ÷ 4M Ω , increasing the X-Tal coupling capacities (keep Qz→Base > Qz←Collector better result in most cases) boosts startup also degrades waveforms , increasing R.e shunting capacities gives better fq. responce below the target fq. (it likely dosent boost startup) , adding C-R chain from base to ground kills some unwanted amplification ? improves waveforms especially I.qz ? adjusts/stabilizes the osc.fq. (it takes a lot of playing around to specify the type of oscillator better - i just found such frustrating at this time) -- the main point -- it uses trivial & flexy range of capacities !!! ((if you dont much change the C8,9,2 - the operating power / other component values  - can be widely modified**))

the prev. v.-s ::
32768kHz Qz-resonator osc. test-dev.
hfhfhf2Bspcfd

32768kHz Qz. Resonator Osc. test/dev.

shortly i need to build a fq. counter coz the 1 built in in KYORITSU KEW1018 . . . "is not too fun to go"
to conduct some IR Comm and other strobed-LED experiments

for that we need a time setting oscillator that's fq. can be fine tuned . . . dev. snap-sh's

---

the LM4250 or К(Р)1407УД2 (actually what i got) seems to be a tricky thing to set up right - it'll take a number of real world tests before an attempt to connect it to anywhere ...

blah blah blah . . .

there's some shit with setting up the LM4250 model ::
.cir - file :: LM4250.cir ◄ .include LM4250.cir
.asy - file :: LM4250_NS.asy ◄ [F2] or Insert New Component
Component - Value :: LM4250/NS ◄ . . . the 3-rd variant of naming your LM4250

2015-08-17 :: the LM4250/KP1407yg2 is actually quite easy to use (you only have to pre-balance/-offset it) if you use a component level model 

[EoF]