TESORO GOLDEN SABRE

[simonbaker]

problem - voltage spikes

I am testing oscillator and negative supply section on a breadboard.

I get quite big voltage spikes (.1 to .4 volt) on V+ rail due to CD4024 counter and push-pull transistors TR7, TR8. Spikes very narrow, couple of micro-seconds, slight ringing at about 400 kHz.

(see attached picture - top trace, tx coil voltage, 5 v/div. Bottom trace - V+ with AC coupled probe, .2 v/div.)

I can reduce the spikes by putting 22 ohm isolation resistors between V+ and component with capacitor from 22 ohm resistor to ground (.47 uF across CD4024, 100 uF across TR7, TR8 ), but spikes still there a little.

Do these spikes exist when parts mounted on PCB, or do they go away?

Thanks,

-SB

[simonbaker]

spikes

I found a major cause of the spikes on the V+ rail.

The negative supply push-pull transistors TR7 and TR8 are both on for a split second as the square wave input to their bases transitions from low to high and vice-versa. Basically, they short the power supply for an instant, causing a spike.

I changed the base divider resistors R11 and R12 to 470 ohms instead of 2k2 ohms so each transistor won't turn on until the square wave reaches it's max and min peaks. This greatly reduced the power supply spikes.

I don't know if the transistors are driven hard enough now to supply good current or are saturated enough to avoid overheating, but it seems to help with the spikes.

I would be interested if someone could put a scope on their V+ and let me know if they see any spikes.

Thanks,

-SB

[Max]

I am testing oscillator and negative supply section on a breadboard.

I get quite big voltage spikes (.1 to .4 volt) on V+ rail due to CD4024 counter and push-pull transistors TR7, TR8. Spikes very narrow, couple of micro-seconds, slight ringing at about 400 kHz.

(see attached picture - top trace, tx coil voltage, 5 v/div. Bottom trace - V+ with AC coupled probe, .2 v/div.)

I can reduce the spikes by putting 22 ohm isolation resistors between V+ and component with capacitor from 22 ohm resistor to ground (.47 uF across CD4024, 100 uF across TR7, TR8 ), but spikes still there a little.

Do these spikes exist when parts mounted on PCB, or do they go away?

Thanks,

-SB

Hi,
they are always there... amplitude can change but you cannot totally eliminate: it's a drawback of design.

Kind regards,
Max

[eclipse]

Perhaps placing decoupling capacitors on every IC would eliminate eventual problems. If the IC you are using don't have high power supply rejection ratio could be a huge problem hugh?

[ivconic]

Friends,
despite all the mods included so far, i am not satisfied at all with audio!
Last mod with monostabile is good enough as it is, yet its main drawback is constant and target (signal strenght) independable last,pitch and delay.
So...here i am back again on the start
Original audio seems to be best so far, better than all mods we included. Why? Simply, operater can take most of informations on target, from it.
I would be more happy if it is more amplyfied and more loud.
One thing also noticed; when speaker 8 ohms is used (without mods) "depth" in air is reduced significantly. At very end of detection, audio start with cracks. For example; coin at 35cm - 40cm producing just "cracks" not pleasant "beep". And even than those cracks are so mum and silent.
When resistor 10 ohms added in serie with speaker, those cracks turns to "beep" very pleasant and fine, but very mum and silent. I have to put speaker on my ear close to hear those beeps.
Closing coin to coil, beeps are rising in amplitude. 15cm away from coil and closer beeps are loud enough and very pleasant.
Damn!
During last year, making over 25 TGSL's so far, (and over 30 coils) i noticed one very interesting thing;
With some coils audio is better and louder, with most of coils audio is weak!?
Does this depend on coil itself??? Other words, depends on signal amplitude?
Cose...i am confused,mostly i make same coils using same datas. Even than some coils "producing" better audio and some worse!?
So.....problem with audio is still here and it is not solved.
How to solve it? How to mod audio arround MPSA13 and make it louder (much louder) but to maintain simplicity and limited pcb dimensions?
Working on it...damn!

[ivconic]

One more thing;
Al foil faraday cage affects phase significantly...some 10 degrees! Due inproper thickness of foil i presume. Yet it is not an issue cose later this can be corrected when nulling. This was discovered by coincidence recently!?

Also...filler mass i am using (HV split TV trafos mass) also has GREAT influence on phase!? Shift it for 5-8 degress!? Still this is acceptable and solveable with GEB trimmer later.
Or adjusting 15nF at RX by adding 1-2nF later..
Also....using ic sockets at LF353 place i discovered too that TL082 "shifts" phase more than LF353. Pay attention on this! Do not be confused when experimenting with various ic's at LF353 place...

I did almost all possible "wild" experiments arround last comparators and last LM358. Changed all the suspectable components...nothing!
Best setup is original TGS setup. Not neither one improvement with other values. Tesoro guys did pretty correct maths there.

Yesterday i discovered one more thing;
at my workshop, right now, i have two coils for tgsl; one for 14.6kHz and other for 11.62kHz. And you know what?
The one for 11.62kHz when connected on tgsl "producing" slightly better and louder audio!?
So i am wandering; are we making huge mistake from a very start?
What is the REAL working frequency of TGS? And what are the real inductances of original coil for TGS?
Might be that band passes are designed so strictly to "press" and filter all above the range? Logically, signal is attenuated, and later stages in detector are geting pretty weak signal than...

Also i never saw schematic and possible differences beetwen TGS and TGS II.
Any chance somebody to post eventual TGS II schematic?

[Max]

One more thing;
Al foil faraday cage affects phase significantly...some 10 degrees! Due inproper thickness of foil i presume. Yet it is not an issue cose later this can be corrected when nulling. This was discovered by coincidence recently!?

Also...filler mass i am using (HV split TV trafos mass) also has GREAT influence on phase!? Shift it for 5-8 degress!? Still this is acceptable and solveable with GEB trimmer later.
Or adjusting 15nF at RX by adding 1-2nF later..
Also....using ic sockets at LF353 place i discovered too that TL082 "shifts" phase more than LF353. Pay attention on this! Do not be confused when experimenting with various ic's at LF353 place...

I did almost all possible "wild" experiments arround last comparators and last LM358. Changed all the suspectable components...nothing!
Best setup is original TGS setup. Not neither one improvement with other values. Tesoro guys did pretty correct maths there.

Yesterday i discovered one more thing;
at my workshop, right now, i have two coils for tgsl; one for 14.6kHz and other for 11.62kHz. And you know what?
The one for 11.62kHz when connected on tgsl "producing" slightly better and louder audio!?
So i am wandering; are we making huge mistake from a very start?
What is the REAL working frequency of TGS? And what are the real inductances of original coil for TGS?
Might be that band passes are designed so strictly to "press" and filter all above the range? Logically, signal is attenuated, and later stages in detector are geting pretty weak signal than...

Also i never saw schematic and possible differences beetwen TGS and TGS II.
Any chance somebody to post eventual TGS II schematic?

Hi,
now your post make me dubts too of frequency, reading here:

http://www.tesoro.com/info/manuals/older/goldensabre/

you'll see there's 12Khz as operating frequency for it... but that's very confusing cause I read in other places docs by Tesoro that talk about 14Khz...

???

Also at golden sabre II you read it:

http://www.tesoro.com/info/manuals/older/goldensabreii/

??????

But at golden sabre plus... :

http://www.tesoro.com/info/manuals/o...ldensabreplus/

15Khz ! Bang! Actually a bit less... 14.7Khz I think. Here we are... we are using the TGSplus frequency but are with TGS schematic... so we have larger caps at colpitts...

What's wrong with 14 KHz ??? Uhm... maybe we have slight phase shifts at disc and geb... due to that, that reminds me at the strange behaviour we get at one pin of lf353... with dumped oscillations.
But also we experience a slight increased performance on very thin stuff... like foils.

I think most of all made till now can be easy converted for new coils using 12Khz but don't know if worth...or not.

I also have thoughts aboud coils... always thinking actual TGS freq was higher than declared with 8'' round CC but seems Tesoro specify exactly 12Khz with it! So it's not cause of e.g. great coil... seems this is the original frequency.

But I see a big problem for experimenting at 12Khz: the frequency divider will give 6Khz not 7.3Khz

You'll think maybe it's not a huge problem but it IS: in bandidoII the big part of voltage converter inefficiency is due to the 5Khz it gets as driving frequency... and could be problematic running TGS with lowered negative rail... unless using really accurate component selection.

What do you think about ?

Kind regards,
Max

[tiktak]

Hi,

Hi,
Max.
So its the operating frequency afther all!So if you think that with 12kHz will be problemes.We can try making lower frequency coils and see whats best but near 12kHz.I see that Ivconic says with 11.62kHz slice better audio.
))))

[Max]

Hi,
Max.
So its the operating frequency afther all!So if you think that with 12kHz will be problemes.We can try making lower frequency coils and see whats best but near 12kHz.I see that Ivconic says with 11.62kHz slice better audio.
))))

Hi,
I think device could work at 12Khz, and even less... but to work good the negative rail must be at a suitable level... and that depends much on current requirements at ics.

Probably using original parts the current needed is low enough not to lower much (absolute value) the negative rail at an unreliable level for safe operations.

So it will narrow much the ability of changing components there for replacement unless using very low power op amps and the like.

I think also that Tesoro made the TGSPlus version running at 15Khz mostly cause of the more stuff the Plus have to supply in its circuit...

Kind regards,
Max

[simonbaker]

Hi,
they are always there... amplitude can change but you cannot totally eliminate: it's a drawback of design.

Kind regards,
Max

Maybe I worry too much about spikes because I saw them in output of LF353 - but now I wonder maybe my LF353 was blown, will check.

Smaller R11, R12 = smaller spikes -- may keep at 470 ohms or a little more and test. I have other ideas to reduce spikes, some diodes and small caps to delay turn-on, turn-off of TR7, TR8 so not on at same time.

Eclipse - I tried some decoupling caps (and isolation resistors) and helps but not as easy as you would think, spikes were very strong. But maybe important, as you say depends on rejection. I think a combination of tricks are worth trying (don't like spikes). Also, maybe C28 is better on output of LM7808 instead of input (or maybe on both).

I want to get finished TGSL because questions of audio, coils, and frequency very interesting for experiments, maybe more improvements possible.

How can coil change audio so much? Very interesting. I hope Ivconic will save two models, one with good audio, one with bad, for comparing. I want to see signal feeding the Darlington in both cases. Good mystery.

Max, I get -7 v or better with no load from negative supply even though I read some people have less -- seems very healthy. If oscillator is slower (12 khz), would bigger caps help (470 uF)? What if we drive it directly from oscillator frequency instead of 1/2 divider?

[eclipse]

OK lets try with lower frequency.
I made today standard coils (255x137 former) - 106w & 100w @ 0.28
but I can make few coils more more test results.
Suggest new coil data for the same former.
Perhaps 85w-87 will be OK for 12 Khz? And what about Rx coil then, we can't use the same coil as for 14 Khz can we?

[simonbaker]

R40 question

R40 is positive feedback on U106b (final LM393).

On older TGSL schematic, it is directly from output of U106b.

On latest schematic (gift pack), it is after diode D12.

Why was it changed and what effect?

Cheers,

-SB

[Max]

OK lets try with lower frequency.
I made today standard coils (255x137 former) - 106w & 100w @ 0.28
but I can make few coils more more test results.
Suggest new coil data for the same former.
Perhaps 85w-87 will be OK for 12 Khz? And what about Rx coil then, we can't use the same coil as for 14 Khz can we?

Hi,
to deal with 12Khz you need more turns on both TX and RX coils.... say you need 10 turns more... so you'll get about 110turns at TX and then 115 at RX.

But I don't tested that so I'm simply guessing numbers now.

Simon: about the frequency I had thought about using directly the oscillator frequency to do that stuff... but think the absence of a squared signal is a problem ... so I think must be some shmitt trigger there... before injecting the signal at transistors base. The 4024 give that kind of output... well squared signal.

Kind regards,
Max

[simonbaker]

Hi,
to deal with 12Khz you need more turns on both TX and RX coils.... say you need 10 turns more... so you'll get about 110turns at TX and then 115 at RX.

But I don't tested that so I'm simply guessing numbers now.

Simon: about the frequency I had thought about using directly the oscillator frequency to do that stuff... but think the absence of a squared signal is a problem ... so I think must be some shmitt trigger there... before injecting the signal at transistors base. The 4024 give that kind of output... well squared signal.

Kind regards,
Max

I agree, need squared signal, means more parts if use oscillator freq, too bad .

Now I'm a little confused about voltage spikes because I found my oscilloscope probe can contribute to spike -- put probe1 on V+ to monitor spikes, touch circuit with probe2 and probe1 shows increased spike! But only true for very small quick spikes I think (hope).

I think the following is useful in general to minimize V+ spike and noise:

1. Move (or add) 220 uF cap to output of LM7808 - reduces ripple from oscillator too.

2. Make R11, R12 in negative supply driver approx .1 X R9, R10 - reduces time TR7, TR8 on simultaneously, less spike.

3. Consider larger R8 to not drive TR6 so hard.

4. Consider larger R7 to not drive CD4024 clock so hard.

CD4024 is weird, can cause crazy spikes depending on connections. Trying to find ways to minimize it.

5. Isolation resistors, decoupling caps on CD4024 and TR7, TR8 driver.

Ok, moving on...

Cheers,

-SB

(still wondering about R40 connection, before or after diode D12..why?)

[Unregistered]

Sorry if its stupid question, Im not strong in teoretical electronics, but if negative supply section is causing so much trouble, isnt just better use some DC/DC converter, like:
http://www.tracopower.com/ ?