Type 284 Pulse Generator tr<=70ps



Fast Rise-Time Generator

The Tektronix Type 284 is a very fast Pulse Generator and comes with a risetime of <=70ps with 50 kHz repetition rate. This repetition rate is high enough for a clear display on oscilloscopes, compared to the approx. low 700 Hz repetition rate of the Type 109 mercury relais pulse generator a big advantage.

The instrument is fast enough for testing Sampling Oscilloscopes or vertical amplifier risetime. Another application are fast pulse in Time Domain Reflectometry or any general purpose risetime measurement. The second output has either a square-wave or sine-wave output.

Square Wave: 100 kHz, 1 MHz and 10 MHz with a 50% Duty Cycle and 1V, 100mV and 10mV amplitude into 50 ohms.
Sine Wave: 100 MHz and 1 GHz with a 100mV amplitude into 50 ohms.

Pulse Amplitude: >=+200mV
Low Pulse Flatness <=3%p-p within the first 2ns.
Pretrigger capability.

The Service Manual has a lot of useful information about this instrument and application hints, it's really worth to search for it. This instrument is seldom available. Sometimes it is selled cheap on second hand markets, but in most cases the price is high. There are not many generators available with such fast risetimes. Unfortunately the shown instrument don't work, it has a fault in the power supply and will be repaired during wintertime.




Einstellbarer Versorgungsspannungsbereich, das Ändern dauert nicht mal 30 Sekunden.

Selectable voltage range, takes not more than 30 seconds changing the supply voltage range.
Two fuses within the holder for each range.









Als Ausgang verwendet dieser Generator noch das GR-874 system, neben den guten elektrischen Eigenschaften zeichnet es sich aus, dass dieser Steckertyp keine Buchsen und Stecker kennt, jede der Verbindungen kann beides gleichzeitig sein, einfach um 90° verdrehen und aus Buchse wird Stecker und umgekehrt. Leider ist dieser Typ nicht mehr aktuell an Messgeräten zu finden.

Outputs came with General Radio GR-874 connectors, which are unfortunately no more build.






The instrument can be slide out of the mainframe.
Very simple construction to reach all parts when reparing or service.
Takes not more than 10 seconds to slide out the Plug-In.



Amazing layout in the sine generator



Quartz and LC controlled oscillators



1ns and 10ns Timing Adjustment



Airline with the tunnel diode mounted inside.





Power Supply Rectifier - somebody already repaired something inside



Pulse Balancing Adjustment - Puls Offset






Diese GR auf BNC Adapter sind fast kaum mehr zu bekommen, der Gebrauchtmarkt ist die beste Quelle.

The previous owner supplied the instrument with two GR-874 to BNC adapters, which are getting seldom nowadays.
They are difficult to get, almost only on the second hand market avialable.





Repair and Calibration (2010)

Repair and Calibration 6. Januar 2010

It's highly recommended to read the Instruction Manual when handling a 284 generator.
The Instruction Manual shows operating details, it's really a useful document.


Fault 1:

The instrument worked under 115V AC, but didn't work with 230V AC, what a fault?





Why the instrument worked under 115V and not with 230V ?

Im 230V Modus ging gar nichts, da eine der in serie geschalteten Wicklungen unterbrochen war, es fließt durch beide Wicklungen kein Strom, daher auch kein magnetisches Feld an den Sekundärwicklungen. Im 115V Modus hingegen liegt eine der beiden Wicklungen an Spannung und daher unter Stromfluß. Bei 115V wird das Gerät nur aus einer Wicklung versorgt, bei der geringen Leistungsaufnahme und der Dimensionierung funktioniert das Gerät trotzdem.


In the 230V mode are both transformer windings in series, with a broken winding, no current can flow. In the 115V mode are both transformer windings parallel connected, if one winding broken, in the other still current flows. The instrument has 6 watts and worked with one primary winding only, the secondary voltage was high enough.



On pin 1 a broken wire, photo shows the wire for repair.

Diese kleine Drahtbücke wieder am alten Draht anzuflicken war wieder eine Fummelei und Strafarbeit. Schwer zugänglich und ultrakurz, nicht mal richtig fotografieren konnte man es, aber es hält.

It was a hard work to resolder the broken transformer wire again, luck that it was not necessary to remove the transformer.


Fault 2:

Im Pulse Modus war die Ausgangsamplitude unstabil, es roch schon förmlich nach einem Wackelkontakt, der überall sein konnte. Manchmal blieb die Amplitude für Minuten stabil, dann wieder chaotisches Verhalten. Der Fehler sah so aus, dass die höchsten Frequenzanteile noch gut übertragen wurden, die etwas niederfrequenteren jedoch gedämpft, das sieht dann so aus wie bei der 1:10 Tastkopf Kompensation wenn bei falscher Einstellung der Peak nach oben schießt.

An mehreren Stellen nach dem Fehler gesucht, letztendlich zur Erkentniss gelangt der Fehler muss irgendwo innerhalb der Tunneldioden Luftleitung liegen. Feststellen lies sich das durch gezieltes Klopfen mit einem Schraubenzieher, aber es dauert ein wenig bis man lernte es gezielt zu provozieren.

The pulse mode had an unstable output amplitude and a high frequency error. In some Time/Div settings, highest frequencies were passed, some lower frequencies didn't passed, looked like an overcompensated X10 passive probe. Sometimes after knocking on the device, the amplitude jumped to a reasonable value - sometimes after knocking the amplitude was wrong again.

I tried many things to find the error, it could be a bad soldering, an internal broken part or just a bad connector. After locating the place of error by knocking with a small metall srcewdriver I could locate the fault. Something must be wrong inside the tunneldiode airline.




Removed airline (please refer to instruction manual for details when removing)

  1. Remove all four SMB connectors (note their position for reconnecting)
  2. Loosen the most left nut, near the GR connector
  3. Lift up and remove the airline from the chassis
  4. Next, resolder the small brown resistor from the right SMB-conncetor on the photo
  5. loosen the right nut, resistor comes out.

Removing is easy, it looks more difficult than it is.






Alle Kontaktoberflächen wurden gereinigt und mit Alkohol gereinigt, anschließend mit einem Kontaktöl gegen zukünftige Oxidation ein wenig geschützt. Der Widerstand bekam auf der linken Seite etwas neues Lötzinn als Kontaktfläche, das alte sah schon irgendwie etwas mitgenommen aus. Pass bloß auf diesen Widerstand auf und mach ihn nicht kaputt.

I cleaned all contact surfaces and put some contact oil for protection. I put a litte bit new solder on the left side of the resistor, I didn't trust that old contact. Be careful with this resistor.





I was lucky that I need not to open the tunneldiode housing.





Dies ist ein 54 Ohm Spezialwiderstand bestehend aus dem Netzwerk R184A, R184B und R184C in einem Bauteil ! Die gezeigte Kontaktfläche gehört zum R184A und bildet die Masseverbindung zur Luftleitung, dieser Kontaktbereich war stark oxidiert, schon richtig halbschwarz, vermutlich war das die Ursache für die schlechte Verbindung. Sei vorsichtig mit der Arbeit an diesem Widerstand, wer den kaputt macht, der steht garantiert vor einem gewaltigen Ersatzteilproblem.

This is a special 54 ohms resistor network, consists of three different resistors, R184A, R184B and R184C in one part ! The showed contact from R184A to the airline ground was very oxidized (here already cleaned), I think this was the fault. I put some contact-oil over the contact surfaces. Be very careful when handling this resistor, it is a special high frequency part, you won't get a replacement.




Tunneldiode D180, R180, C185 and C186 are mounted inside the tunneldiode housing.


Pulser worked again after reconnecting the airline !


Checking the Power Supply


The 284 has approximately only 6 watts and two regulated DC supply voltages, +20V and -20V.

   
100 Hertz Power Supply Ripple on the Rectifier Electrolytic Capacitors.
They carry a 40V DC voltage, they remain in the circuit, their ripple is not too much.



C266, 100µF Electrolytic Capacitor on the regulated -20V supply
The spikes and ripple caused by the the 50 kHz pulse mode, there is no 100 Hz Ripple observable, rejected by regulator.
C266 and an paralleled 1000µF capacitor. Additional C reduced the remaing switching ripple.
C226, 100µF Electrolytic Capacitor on the regulated +20V supply
The spikes and ripple comes from the 50 kHz pulse mode, there is no 100 Hz Ripple observable.
C226 and an paralleled 220µF capacitor. Additional C reduced the remaing switching ripple.

Der Spannungsregler kann die 100 Hertz Brummspannung in exzellenter Art und Weise unterdrücken, auf dem Oszilloskop ist nichts mehr davon zu sehen. Bei den hohen Frequenzen streicht er allerdings die Segel und unterdrückt nur noch gedämpft. Ich habe als Parallelschaltung zu den alten Elkos noch Folienkondensatoren ausgetestet, deren niedriger Ersatzserienwiderstand ist keine so gute Idee, der Spannungsregler antwortete prompt mit einer kleinen Oszillation, auf eine Vergrößerung der Kapazität mit Elkos reagierte er angenehm.

The voltage regulator rejects the 100 Hz noise excellent, but the regulators can't reject the fast transient loads in the same way as the 100 Hz. These added electrolytics improved the supply. I tried also low ESR foil capacitors in parallel, the regulators started to oscillate with them.


1000µF paralleled to C266
.


Calibration Procedure taken from the Instruction Manual

Article shows not all steps in detail, please refer to the manual.

Step 1. Check or Adjust +20 Volt Power Supply


Adjust R227 for a reading of +20 volts +/-0.2 volt.


Step 2. Check or Adjust -20 Volt Power Supply


This -20V supply get the reference from the +20 volt circuit. Reading for -20 volts +/-0.6 volt.

Step 3. (Optional) Check Power Supply Regulation and Ripple

see above.



Step 4. Check or Adjust Square Wave Amplitude, R51



A precise 50 ohm termination resistor from Switzerland.



Ready



Remove Transistor Q35 from the circuit. This will
disable the square wave and the instrument outputs a
DC voltage equivalent to the square-wave amplitude.

This kind of circuits allow a calibration with DC voltage equipment only and simplifies calibration and the requirement for demanding test equipment.



This was the 1V setting before calibration.






The settings after adjusting R51 for the three ranges

1V
100mV
10mV

The voltages can't be setted for each range individual, R51 sets all of them.
Try to find a good setting for all three ranges.
You are not satisfied? - than take a soldering iron and replace the resistors in the attenuator.


Next test verifies the square wave switch, output voltage under switching.


For this test set the 1 volt DC exactly to the center horizontal graticule on the CRT.
Don't move the offset potentiometers for the next steps, leave the Y-amplifier good warm-up before proceed.



Put back transistor Q35 in its sockets, the square wave appears on the CRT. The flat shoulder of the square wave should not move away from the former beam position (when DC), movement should be within 20mV.

Failed

Ein blöder Fehler, die Reduzierung der Amplitude beim Schaltvorgang ist etwas höher als erlaubt. Wahrscheinlich ist irgendwo ein Signalpfad zu hochohmig, mag es der Transistor oder eine Steckverbindung sein. Ist allerdings nicht so wichtig, der Rechteckgenerator hat außer der Frequenz nichts mit dem Pulsgenerator zu tun, und genau den benötige ich. Irgendwann mal ist dieser Fehler fällig.


More than the allowed 20mV movement below center graticule, instrument need a check. May be in the switching transistor path there are somewhere too much ohms. But it doesn't matter much for my instrument, I need the 284 pulse generator mode.

Next test, 100ns time period amplitude check.


 
Typical display of the 100ns time period amplitude check.



Step 5. Check or Adjust Sine Wave Amplitude (10 ns and 1 ns Periods)



Check - The peak-to-peak amplitude should be 100 mV +/-20%.
Adjust C66 if necessary (instrument was in specification).
Using a S-4 Sampling Head



Check - the 1 GHz sine wave peak-to-peak
for a 100mV +/-20%.
Within specification, shows +20% - C72 1ns timing Air-Capacitor should be adjusted.
Be careful when adjusting, C72 area is very sensitive against parasitic capacitances.

Need Recalibration


- Done -
C72 adjusted for exactly 6 Divisions, measured with a the S4.




Step 6. Check or Adjust Square Wave and Sine Wave Periods


100 kHz frequency before calibration.



after adjusting inductor L4 in the 100kHz oscillator.


1 MHz oscillator, no adjusting for L14 necessary.


10 MHz period, not tested - 34401A can't count 10 MHz, no other counter at this location.




Step 7. Check or Adjust Square Wave Symmetry



 
Frequency and 10µs Square Wave Symmetry OK, duty cycle 50%

Das sieht verdammt aus nach einem erhöhten Widerstand innerhalb des Signalpfades, was auch den Anstieg der Flankenzeit erklären würde, da der Widerstand zusammen mit einer Kapazität am Ausgang eine RC Zeitkonstante bildet, die bei größerem R länger zum einschwingen braucht und die Anstiegszeit verlängert.

Rise- and Falltime to slow, don't know why, may be the same fault as in step 4? Increased RC ?

FAILED





Frequency and 1µs Square Wave Symmetry OK, duty cycle 50%
Rise- and Falltime to slow, don't know why, may be the same fault as in step 4?

FAILED


Step 8. Check Square and Sine Wave Trigger

OK


Step 9. Adjust TD Bias

Read Instruction Manual


Step 10. Check Pulse Repetition Rate


Check Trigger Frequency - 20µs period



Check Pulse Period for 20µs +/-2µs


Step 11. Check Pulse Width


Check Pulse Width between 2 and 4 Divisions ( 1µs +100% -0%)


Step 12. Check Pulse Amplitude

Check for a pulse amplitude of >=200mV

Step 13. Check Pulse Trigger Amplitude





Check for a pulse trigger amplitude of 200mV +/-40mV


Step 14. Check Pulse Trigger Width




Check for pulse width >=10ns at the 50% amplitude points.

Pulse shows a ringing in the falling edge, I don't know why.
but it doesn't matter when triggering to the rising edge.
Unsafe when triggering on falling edge.


Step 15. Check Pulse Trigger Risetime





Check for Risetime <=3ns


Step 16. Check/Adjust Snap-Off Current; 5 ns Lead Time, R174






Adjust R174 for 5 ns +/-5 ns between both leading edges
(note measured with a 7A29, compare different risetime between amplifier plug-in and the faster sampler in Fig-5.16).

Do this measurement in the 5 ns and 75 ns position.




Front Switch set to 75 ns, check for 75ns +/-5ns between both leading edges



Internal Switch (new series only) set to 150 ns, check for 150ns +/-7ns between both leading edges


Step 17. Check Pulse Output Risetime and Aberrations


Using a S-6 Sampling Head with 20cm airline, 1ns semi-rigid coaxial cable, SMA to GR adapter, 50 ohm GR terminator

Check for Risetime <=70ps

Check Front corner aberrations should be <=+3%, -7% for a total of <=10% peak-to-peak
after 2 ns the aberrations should be <=2% peak-to-peak
Check back terminator aberations should be <=+6%, -6% for a total of <=10% peak-to-peak.


I don't have a S-6 Sampling Head on this loacation
I don't have the high quality 1 ns rigid coaxial cable

This was the last step in the Calibration Procedure



Some Transient Measurements:

1. Test

2. Test


Using a S-4 Sampling Head (tr <= 25ps)
Two adapters only, shortest possible connection.
  • SMA to BNC
  • BNC to GR-874

Increasing line by adding a 10 cm airline
  • SMA to BNC
  • BNC to GR-874
  • Airline 10 cm
Without 10 cm airline
Displayed Risetime 50ps
Displayed front corner aberrrations +/-5%
With 10 cm airline
Displayed Risetime 50ps
Displayed front corner aberrrations +/-5%
(note: ignore trigger instability in the right photo, this is a matter of adjustment only)

Test - Aberations and Risetime with a Sampling Head and different cables


Using a S-4 Sampling Head and a 7T11A Time Base


Different Test Configurations:

10cm Airline - 50 ohms



30cm Airline - 50 ohms



20cm Cable - RG58C/U - 50 ohms



30cm Cable - RG58C/U - 50 ohms



2500cm Cable - RG58C/U - 50 ohms



60cm Cable - RG59/U - 75 ohms





Ein Sampler System ist von Natur aus ziemlich unempfindlich gegenüber einer Übersteuerung der Verstärkereingänge. Der Sampler kann daher mit der Volt/Div Einstellung nahezu beliebig agieren ohne dass es zu einer Übersteuerung kommt (siehe Kalibrationsschritt Nr.4 Fig. 5-3 und Nr.17), dieser Lupeneffekt kann hier genutzt werden. Man beachte aber diese Aussage nicht damit zu verwechseln, dass ein Sampler keine Überlastung kennt. Selbstverständlich darf der Sampler nicht mit zu hoher Spannung am Eingang belastet werden, ein Blick ins Manual zeigt die Grenzen, leider sind die Sampler nur gering gegen höhere Spannungen ausgelegt - Vorsicht.

Die Zeit für die Erholung eines Verstärkers um aus einer Übersteuerung wieder herauszukommen ist ein selten spezifizierter Parameter, bei einem Echtzeitverstärker sollter dieser Betriebsmodus daher vermieden werden, d.h. der Strahl sollte sich nicht zu sehr weit außerhalb des Bildschirmes befinden.

A Sampling System is inherently immun against being overdriven. For showing signal details (see step No.4 Fig. 5-3 and step No.17), it is therefore valid to set the
Volts/Div. position to almost any desired sensitivity. This is an big advantage of sampling systems compared to real time amplifiers with their limited overload recovery time. Nevertheless, be careful and don't destroy your sampler, you can not overdrive a sampling system with a higher input voltage than specified.

In comparison a real time amplifier system should remain close within the displayed CRT limit, do not overdrive your vertical amplifier when doing amplitude and time sensitive measurements. Amplifier overload recovery time is a seldom specified parameter, do experiments or better avoid.


When setting the sampler to 20mV/Div., then 1 Division is equal to 4%.
 
Shown aberrations remain under 5% when using the fast airlines.

Using different airlines and cables change signal propagation time (not shown), trigger always readjusted for visibility, ignore timing details when the measurement triggers. A longer airline or cable will increase risetime, will shown in the second section. An 60 cm 75 ohms RG59C/U cable will show effects of wrong line impedance. The 7M13 Readout Plug-In indifies all measurements.

Comparing Aberrations:


using a S-4 Sampling Head

Highest frequency contents when using airlines.
Result differs in front corner aberrations.
Back termination damped with long cables.

Start a Aberrations Slide Show (6 photos),
(download exe file, scroll with mouse wheel or right left buttons)



Comparing Risetime:

using a S-4 Sampling Head
Fastest displayed risetime when using the 10cm airline.

45ps

Risetime increase with length and dieelectric cable properties.

In assumption the S-4 has a 20ps risetime, the 284 risetime can be recalculated:
Sqrt(45*45*ps*ps - 20*20*ps*ps)
approximately 40ps

This risetime corrosponds to approx. 8.75 GHz Bandwidth (Bw=tr/0.35)

Start a Risetime Slide Show (6 photos),
(download exe file, scroll with mouse wheel or right left buttons)





Test - Aberrations and Risetime with a Real Time Amplifier and different cables



using a 7A29 Real Time Amplifier with 1 GHz bandwidth
together with a 7B15 timebase in a 7104 mainframe.

You will see effects on risetime and aberrations of different cables, 50 ohms and 75 ohms.
Measurement series done with two different 7A29 for a better comparison.
Both amplifiers with unknown calibration status - as is.

Left Plug-In Amplifier Serial No: Bxx995
Right Plug-In Amplifier Serial No: Bxx361

(noting serial numbers for my personal records only, for a later comparison).


   
Two 7A29 in a 7104 for comparison, drivem by the 284 with the 10cm airline.

Displayed Risetime Serial xx995:    340ps
Displayed Risetime Serial xx361:    330ps


Start a Risetime Slide Show for Serial xx995 (12 photos),
(download exe file, scroll with mouse wheel or right left buttons)


Start a Risetime Slide Show for Serial xx361 (12 photos),
(download exe file, scroll with mouse wheel or right left buttons)


Before making a photo, as a rule two waveform points were held always constant:


Results

  • Airline cause fastest risetime, due higher frequency contents.
  • Difference between 10cm and 30cm airline can't be seen on a 7A29, this would require the higher frequency bandwidth sampling system.
  • Risetime increase with cable length.
  • Amplitude decrease with cable length.
  • RG59/U risetime adequate for the 60cm length.
  • RG59/U impedance mismatch due mixing a 75 ohm with a 50 ohm system. Toggle between the two photos 10 and 11, the RG59 shows a significantly lower shoulder compared to the 30cm cable with 50 ohms within the first 6ns. These 6ns seconds caused by the progation time for this 60cm cable.
  • 20cm cable shows the behaviour looking slower than the 30 cm cable, this is caused by a non-constant BNC contact. Depending on touching, pressing, pulling on the BNC connector, risetime and ringing can be effected withing the first nanoseconds. Therefore for higher frequencies use a mechnical more constant system, like SMA, N, GR .....


Conclusion

Der 284 ist ein tolles Gerät um schnelle saubere Rechteckflanken zu erzeugen, dazu noch handwerklich prima gefertigt. Seine Fähigkeiten erlauben es Oszilloskope (alle mir bekannten) auf eine schnelle Rechtantwort hin zu kalibrieren, er ist ausreichend schnell um viele Sampling Systeme damit ausreichend testen zu können. Die Anwendung in Kabelradar Anwendungen ist ein weiteres wichtiges und großes Anwendungsgebiet.

Das Heranarbeiten an Hochfrequenztechnische Problemstellungen fällt mit diesem Gerät wesentlich leichter, insbesondere entsteht damit ein echter Bezug zur Realität, weg von der schnöden alleinigen Theorie - wie oft schon haben Generationen von Elektrotechnikern (meist nur in ihren Studienzeiten) seitenweise mit Impedanzen und Reflektionsfaktoren herumgerechnet ohne eigentlich nicht wirklich zu wissen was sie da tun. Es wird gerechnet bis zum Sankt Nimmerleins Tag und nur selten der reale Bezug hergestellt - ein Trauerspiel. Die Erinnerung ist immer entsprechend vergesslich, abgesehen von den Berufs- und Amateur Hochfrequenztechnikern.

Rechnen können viele, das ist nicht schwer, manche Formeln herzuleiten oder die schwierigen abzuschreiben auch nicht, aber sie zu verstehen oder gar was funktionierendes daraus zu machen das ist ein ganz hartes Brot, das geht leichter über eigene Erfahrungen. Natürlich den Hut ab vor hyperkomplizierter Mathematik - keine Frage - diese hat sich aber in der real erfassbaren Elektrotechnik hingegen als nur schwer realisierbar oder oft ziemlich gegenstandlos erwiesen; wer sie bei einem Problem nicht durch andere einfachere Lösungen zu vermeiden versucht macht schon den ersten Fehler - nur wenn es unbedingt so sein muss.

Mit solchen Geräten hingegen kann es vorkommen dass man sich drei Tage am Stück damit beschäftigt um ständig seine Gebiete zu vertiefen. Eine ist eine wahre Freude 40 ps Impulse oder die Unterschiede zwischen einer 10cm und 30cm Luftleitung live erlebt zu haben.

Solche Geräte (wenn nicht alt dann neu) sind eigentlich ein Muss in jedem anspruchsvolleren Elektronik Labor.

The 284 is a nice source for testing fast transient response of amplifiers, sampling heads and a source for TDR, Time Domain Reflectometry measurements. Very helpful instrument for learning and working with high frequency.



Additional

(11. Januar 2010)
Today I tried to fix the square wave amplitude problem


J55 is the square wave output going to the attenuator. Q35 and Q55 is a fast differential amplifier in a comparator mode. Q55 base DC voltage set the output amplitude, Q35 base receives the AC signal frequency.

Q35 and Q55 emitter held almost constant voltage, therefore a constant voltage across R38 and R39 cause a constant current. This currents flows current flows either through path R40 or R58 depending on the alternating Q35 and Q55 base potentials. The result is a square wave voltage across R58.

Q35 and Q55 are responseable for the waveform quality. They are type 2N4258, this a ultra high speed switching transistor with a 700 MHz transit frequency. Q55 has to be a perfect fast switch, the DC Beta should be similar to the Beta under switching - if not - base current and collector current through R58 will will be different under AC and DC, resulting in a different DC and Switching mode amplitude.

Q55 had only hfe=32 @ 50mA and Uce=1V, the datasheet specified hfe=min.30 under the same conditions. This transistor reached the lower hfe limit, this part getting old in a long life of operationg hours under high currents. Q35 and Q30 were also in a weak condition. A curve tracer is a very helpful tool.

Q35 determines the waveform quality, especially for the squarewave risetime, but is not responsible for switching amplitude errors.

What to do now?


Unfortunately I don't have a new original transistor replacement on stock, I took another transistor and tried different types for Q35 and Q55.

2N3906 and BC Types - bad results - 100 ns Squarewave looked funny,  transistor switching times seem to be too high or capacitances too high, I don't know in detail.

2N2907A - very good for the 10µs and 1µs low frequency square waves, due their higher beta, the rise time decreases a lot !, but the 100 ns Squarewave looked funny too.

Now trying all other 2N4258 in the circuit (Q40 and Q30) on Q35 and Q55's place, taking the best one.

Now I did experiments with different other transistors, the best overall result reached by replacing the Q55 with a 2SA1085 Type, the switching amplitude getting closer to the DC value and the waveform improved, unbelievable with such a slower type. Only the risetime decreased a little, but this doesn't matter, a flat square wave shoulder has more practical value, the former risetime peak almost disappeared. It was luck to find that type  - it is a low noise, low speed transistor with high beta - high gain is good for amplitude accuracy - and it seems this transistor capacitance fit good in this application. At first glance theory it is the wrong transistor - but here the part worked !



Changing transistors.


with new transistors
(DC level when Q35 removed)
with new transistors
(Q35 replaced in it's socket)
Reduced by 10mV, with the old transistors 20mV.

- improved, within specification now-


with old transistors with new transistors
waveform and risetime remain unchanged

- risetime is still an fault -
with old transistors with new transistors
waveform and risetime remain unchanged

- risetime is still an fault -
with old transistors with new transistors
less overshooting, more flat

- improved waveform -

in detail
with old transistors
Trigger
with new transistors

Trigger

- trigger falling waveform improved -

Q35 ==> Q40
(Q35 changed with Q40)

Q55 ==> 2SA1085
Q40 ==> 2SA1164
Q30 ==> 2SA1164


Amplitude Calibration,  I canceled out the 10mV amplitude switching loss, by increasing the adjusted DC by some additional mV, square wave amplitude seems to be correct now compared to other generators. This experiment took me three hours of testing different transistors, it not important what I did here, but it was a interesting for me.



www.amplifier.cd        Tektronix Gallery        Index

www.amplifier.cd

Impressum und Haftungsausschluss