|K5JXH , Project,
rebuild one old 1986, Wavetek 3006 Signal Generator. "MY SIG GEN "
This is my Journal for R&R, rebuild & repairs. (finding one complete and not hacked up is a joy)
It was lots of fun , making 1980 gear (instruments) come back alive and very useful in the ham shack. (for sure tuning and testing rigs)
Sure the HP 8656b is better , but costs 20 times as much !(that is if the HP plastic gears are not broken or about to , with cracks)
A side note:
I'm building precision voltage standard now (on digital and one fixed ref. 10.000v) and have chip that does 10.000MHz, TCXO, ± 1ppm from 0°C to +50°C and ± 1ppm / year drift and then trimmed to WWV.
Sometimes the badguy puts (not mine) in logic chips that have silver plated leads into sockets.
Bad this is, as the maker of the chip and his data sheet cleary says no Silver opted parts are to be used in any sockets (they must be soldered in)
The wrong chip pin leads look jet black?, BINGO. (looks like 100year old mom's relic , real silver spoons; never once cleaned.)
Do not run the DPS2(PSU) at full normal loads and loose or unbolted from the case or it will OVERHEAT. (but you can if you drop off the P1 load connector and only have AC side connected)
The truth is ,that humans can fix it, after all it is NOT full of SMD chips with 1/8th inch sized chips & bottom solderballs, and resistors small as a flea.
Yes, a human can in fact see the parts, and remove them lacking a microscope.
It's not full of PCB's that cost $1000+ each like say a newer scope has. yah.
Some call this a boat anchor, sure it is smaller that TUBE GENs (HP 608s) and way lighter, but will not fit in your shirt pocket nor cost you $3000 to buy one. (it is heavy, but is compact , in my book)
This GEN (SIG-GEN) is VCO based GEN with PLL controls, for dead accuracy on frequency and has regulated levels. (means you change Freq, and the output levels hold where you set them)
The noise is good, here, on the RF jack, only phase change noise (PLL), not that horrible DDS , quantization noise and harmonics.
The internal signals are as high as 1440mHZ, 1.44 giga hertz. and takes skills to measure. (if needed at all)
The reason for the DAC and VCO is to make the VCO levels match the non linearities of the Varicap tuners.
Below 40MHz the GEN , using TTL digital division to produce, signals.
The four basic PLL 1 to 4 must be made to work correctly in that order.
The down convert board M115, works (counts down) when the MHZ digi-switches are at 000, this is good test too. 1 to 999 kHz output frequency. It just mixed 10MHz reference to get the lower frequencies.
This board can be removed to get 1 to 520mHZ usage. (pin1 to 9 jumped)
The book cover all theory of operations very well.
The PSU (DPS2) power pack has 2 (18v) LM395 devices, that limit current to 1amp and if overheated ever, the shut down, this is a very very good design. (remove shorts first on loads) 10v drop across this device all the time)
Next up is PSU , LM317 device for 7.3v raw to all 10 total, 78m05 regulators, found inside many modules.
The 317 is a full blown regulator good for 1.5amps (with proper Heatsink, yes ours is) It too has current overload that trips at 1amp hot or 2amps cold.
It too has thermal overloads, trust it to do its job at the silicon die level, I protects it self 2 ways current overload and heat. (always assume DPS2 is ok and modules are in fact shorted , from tantalum caps.
Day 1, checks. (a list)
Punch list (failures) , my chips are date marked 1977 to 1983. (as are pots) Below is my linear progress, power is alway first in all electronic systems. (and modes that fail or not next)
The PSU connector color codes from right pin1 to left is per below. (DPS2)
Pin 1, 27v dark gray
Pin 2/3 = +18v Orange wire reads 0.2vdc (module 12 shorted),
Pin 4/5= -18v Yellow , 0v, found 5 modules shorted, all are corrected now. PSU shuts down for all shorts lower than 27Vdc, 27 has fuse.
Pin 6 , N.C.
Pin 7 = +7.3 raw, = Green large pin 3 (mine is faded to gray on top , but green on bottom pins. mine reads 7.33vdc
Pin8 , N.C
Pin 9, Ground. light Gray wire.
The supply rail +7.3v goes to at least 10 cards, and +18v and -18v to all cards with analog devices. 7.3 must never be below 7v or the regulators go off line.
(the M115 is the worst for shorts, it has 9 Cap.'s inside the can short easy and often, I had 2, tantalums fail here and one 100uf blew it's lid)
All 10 modules have caps that short on card, before the 78m05 regulator even sees 7.3v. I'd pull the front panel connectors first (most easy) and it too had shorted caps +18/-18v rails.
I had no 5v/7.3v shorts.
The 78m series spot regulators on most cards,only does 1/3rd of the current max. a regular 78 series can do. 500mA is max current with ours. (my guess 100mA max and for sure less)
There are many methods to find cure shorts.
Once the bad module is found, it has schematic and the path is obvious and easy to find shorts, pulling each part on the bus shorted, 1 by 1, ( 10uf +caps first then ICs in sockets and last diodes)
Bench testing the module, with a bench PSU. (V/I adjustable)
On each card for example -18v pin shorted I found Teflon wires to a pad there and lifted them all and found which card inside was shorted, when found the end that shorts, I check the 10uf first , one end lifted, bam 1ohms cap shorted.
Omg, there are 7 bad.
OK I have full power now. and RF works perfect from 100khz to 520MHz, and the powersupply noise is less than 5mV (tiny)
Last is fixing bad modulation and bad meter actions in internal modulation mode, (easy fixes both) Audio and DC are always easy fixed, not RF.
Secrets: (and Calc's)
Each PSU rail or line, is limited to about 2amps (both 18v rails or about 20watts max) and is power and overheat protected. The 7.3v rail is limited to. 1.5amps 20 watts (below 150ºC) all 3 run a 10vdc drop.
Each PSU line is below 5mV noise. (scoped out at 100MHz.)
But not raw 27vdc it's full bridge, pulsating DC. 120hz. as expected . (as seen on that huge fuse in DPS2 , +27vdc) the noise is totally limited to the 1500uF cap there and the 1amp flow, at 5.5 ripple full wave. C1 and C8
Both my C1/8 caps MEASUIRE 2v peak to peak ripple. (keep mind filter caps are spec'd. as high ad +80% so 1500 is can be 2700uF. (see?) Mine are date coded 1981 Sprague.
My 10,000uf measures 600mV peak to peak.
C10s ripple will be , at 1amp, is 0.8v ripple 800 mV. max. It will always be less if load is less than 1amp. (for all 3 caps huge) The 3 regulators have no trouble cleaning up raw ripple inputs.
Ripple math, delta V = ripple, F= 120Hz, C = 10,000E-6 (in farads) , i = 1amp. for C10 so 1/fC is the answer.
The regulators in DPS2 (PSU) are both current overload protected and overheat tripping. (trust them to do what they do and that is shut down when shorted rails happen, like mine did and no harm done.)
The AC line current read 0.6 Amp.s or 72watts (VA) This is normal.
A Crude factorial:
That equates to DC side losses + load , like this, the regs burn worst case (3 at once) 10+10+ 7w) or 27watts just in the regs. So the the load burn is (18wx2 +7.3w) 42watts at full load (42+27 =69watts) at full load.(matches)
Seems mine runs near full load and the Heatsink does get WARM. (some heat in the transfomer is real with hand placed on it , 85% efficient is common.) So 61 watts land on the load, not 72.
So my loads are running 88% of max.
Mine runs overnight and does not self shut down from heat or overload.
Room temp. 77ºF (spring day (March), AC is off line nice days)
The hottest reg was the top one at 105ºF , the rear hot spot, facing rear left is 103ºf (the heat sink transfer points) hottest module M10 at 93ºf ,M9 , a close match. Transformer hot spot 92ºF
Each reg. TO-220 has thermal pad that must work and the heat-Sink bar has one too to the rear of the case, the regs must not run hotter than the rear case hot spot or those thermal pads are NO GOOD.
The 2 screws on the heatsink bar facing rear , left, must always be very tight, !!!
I have all the tools to measure current, but cutting up the harness to do that , I pass.
The above facts are plenty accurate enough information to diagnose shorts. (ballpark)
Rear Jack TP pin 36 is 8v peak to peak AM modulator on,at (400 or 1000Hz) on my scope.
History and how it relates to finding parts today and substitutions of parts.
First off the correct part is a Wavetek part that is not sold now, all others are substitutes if the real part numbers do not match up, exactly.
The digital logic classic TTL parts are all to be found, (all means 99% effective) 7400 class chips and newer, the one 93L01 chip is a sub to 74LS42 but rewired ! In fact some can be upgraded to lower power parts.
This SIG GEN run ±18v (36v) rails, one can drop that to 17.9v and be in spec... your choice that, for those many analog chips.
Finding the military version today is impossible, the ones rated for ±22v (44v) but not all is lost, there are other choices.
The ua741A analog IC (OP_AMP) is made by even mighty TI.COM, and sold by Mouser.com (the "A" prime grade version is best, 22 volt rails certified)
The Dual OP_AMP chips RM4558 (@4MHz) and MC1448 (it's just a dual 741 device) But again the VCC needs to be above 18v. (36v rails) Mouser.com sells the RC4558,741A and TL082.
Those 1458 chip to are limited to ±18v, it too is just dual 741C mimic.
Same with TL082 chips, to are the wrong voltage from power and used in this SIG GEN or the TL072, (the 2 on the end means dual 741 form factor) TL = Texas Instruments Linear divison, and is still made !
A new substute I found for all duals in the GEN.
I then found a modern chip that does 44v (±22v from "ON Semi" maker, called the MC34072AP (sold buy good old Mouser.com for a buck) this is a very good chip, a super op.amp.)
I see only one true weakness in this SIG GEN, (age sure) and that is the Electrolytics. No do not replace all 100 of them, that be very hard long work (weeks worth?)
I found that I had only 6 tantalum caps shorted on many modules, and one was , open and 1 blown electrolytic, 100uF on the M115 card.
The only IC failures on mine was the modulator board. IC3
Keep in mind we fix these 1 part at at time; a serial process of testing and repair, then if part can not be found ask for help. Some parts are just not available now.
HISTORY of Semiconductor devices (1970 to now) (do not forget to read page 215+ in the manual on upgrades and parts subs)
Now a lesson in chip history (all types active) be it ICs or diode or transistors.
The key here is not look so hard at one partnumber, the trick is look at the class of device, to see if there are sisters to it , and in fact most times better grade parts ! (for sure today)
The caps and resistors are child play to replace, and are all sold today. (With caps, you can use larger uF and for sure higher voltages, most are for brute DC filters (that blow up,short or leak or dry outside))
All semiconductor chips then and now and ,diodes and transistors are made , then tested and Classified and binned. (bin 1 to 16 grades , then custom numbers assigned to them)bad chips are crushed later.(avoiding gray market junk)
The prime parts are sold for the highest price, Just like your fastest Microprocessors from Intel or AMD are graded.
If you look at say one makers diode chartes, showning all they ever made you learn fast what you can use.
The Diodes may be graded from PIV and reverse leakage, (or in the case of zeners are graded by the zener knee voltage and one part-number assigned per knee voltage)
Other Diode are graded by, it's very low forward voltage drop (Schottly) or even fast recovery devices, (speed and cap loading in pF)
Some makers have like 20 diodes in one current (amps)rated class, and only one is 1Nxxxx listed the others are sold by the makers custom assigned numbers. See all these from Vishay semi, 1amp.
The same exact words can be said by transistors, 2Nxxxx, like this. check out the many JFETS still made and still sold. (see all those other numbers?)
In most cases here all ours are simple general purpose JFETS , except the DUAL JFET.
Finding germanium anything now is hard. (mostly a pre 1970s thing) To see a very old list of parts, check out the 700 page document, (vast types)< shown just to scare you!
The same holds true with Silicon transistors (easy to sub below say 100mhz rated aparts), lots of side grades around the JEDEC 2Nxxxx (we have some 2n5xxx parts and some up around 1.4 GHz. so be careful here)
In fact , many are available off the shelf now and cheap unlike back in 1985, back then you needed to buy parts for a major distributor with a $100 minimum order. ($100 was big money then ,huh?)
Learn that some are sold now in SMD package only, you can convert to leads. (tiny transistors we can do that easy)
JEDEC (2Nxxxx or 1Nxxx) class. (eg. 2N2222, or 1N4001, common as nails)
Many parts are still made ,but not with through-holes ,like in DIP form but now only in SMD ,so we have adapter cards now to make that work. SMD to DIP (dual inline packages)
Military , and tested wild temperatures and even hot cycled and burned in tested at the same times. Mil part are always good choice if matched. (but rare for old parts not made now)
A diode may be sold 3 ways, in different qualities or ratings, and as JEDEC 2Nxxxx or Military grades or custom manufactures grades.
The European System(ES) (check this out see both the 1N pref and the ES system of named parts)
The secret to this is to read the (find) the old original data sheet and behold the devices related sisters there,, (just like 1N4001 a 4002 is better and works prefect)
eg CR101 HP 5082-2800, HP shows its a 1N5711 , see how I did that?, at HP searched? In most cases you will find related part,or one that is better with a different part number.
Why Wavetek calls out HP part is a wild story untold. (my guess is HP sold them 1 by 1 but the distributor only did $100 worth, only. , always a pain in then 70's and 80's back then.
Each case is different, some parts can be upgraded, but in the case of VariCAP , it must match perfectly. MPS3702 (but is sold as SMD now, so... it subs)
Many transistors can be found or upgraded or even with same part-number but the package code is wrong, so you adapt it. JFET's and MOS(IG) are easy, but old germanium transistors are hard to find now.(or impossible)
In the original manual the 11C44 chip is covered for upgrading to 4044P. (see manual below and my subs list below.)
I attempted to find some chips below. (text file)
Know too that if a chip was made in 1986 and you read a 1986 data sheet it may show many versions of the same basic part, or not, then if you go read newer data sheet you many even see a newer number listed, even better parts.
Diodes with higher PIV spec. are in the list in most cases,and work perfectly. (like 1N4001 but you have some from 4002 to 1N4007 in your spares box , go for it)
In some cases the older part is dropped off the list and only the newer part was sold, then next year with a new part number, and is a better part. (for example Diode PIV spec)
Last of all, with the full spec in hand and not getting any where fast, search by function for other makers of parts, that are the same or even better or just pins swapped.
The device may even work with a more simple part and cheap, for sure in the audio modulator most parts there are a very easy substitution, the exact usage of the part matter, and in many cases, substitutions can be easy
The show stopper is logic, if it is not made now and there is no newer chip made for that logic, you can only hope one pops up on ebay.
end swaps and subs.
See my Parts substitution page for my SIG GEN, a simple text file.
The M31a module can be upgraded to M31B, with (2 resistors (100ohm) and 4 newer chips ,75ls112,74ls11, and Tl082, MC4044 ) (but only if any of those devices fail)
My horsed up manual.(50 mega bytes large !) See my new page 79 to see all paths of signal by Jack numbers.
The PLL magic happens here, the Fairchild 11C44, phase detector , this ramp seen below causes the generators oscillators to hold phase. There are 4 to 5 PLL here.
(PLL = Phase locked Loop) There is newer chip substitute covered in the manual, using , new 4044 chip name. (old 11C44)
This chip is the heart of the machine, as are the VARICAPs.
It means that the PLL system holds the dialed in frequency exactly, while set to CAL mode 0.001%) And the front panel LED flashes if any of the 4/5 PLL fail. (remove case lid, to see which LED fails inside for "Unlock")
The up to 5 PLL unlock sources are all "OR" logic tied to the (and flasher) to the front Vernier LED. If any PLL fails the Vernier LED flashes. (in CAL mode)
There are DC OP_AMPs that monitor this ramp below and if out of range, the PLL module glows its LED (on top if it) telling you the PLL has lost phase lock. (or someone played with the monitor pot's (window detectors)
Parting shot, the manual actually tells you that doing the calibration can in fact cure the PHASE lock failures, my advices is to RTM first. (do the CAL steps)
90° Shift of the sin wave = π / 2 radians. So you can see that 90degrees shift can almost be corrected. 1.5 radians sure. (and locks you on phase, at 40mHz crystal accuracy , refr)
And last of all the missing page in the manual telling the TECH. 3 things, what "Wn" cable fits where" what Jack and last my creation, what is the "Jn" number. I added this to my PDF manual.
Mine runs like new now. ( I finished the calibration too)