BJT VBE auto adjust?

[Klaus Kragelund]

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

 

[Clifford Heath]

On 16/4/20 8:24 am, Klaus Kragelund wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust

Find an LED that has the reverse temperature dependency and derive bias
from it.

 

[Klaus Kragelund]

On Thursday, April 16, 2020 at 12:58:32 AM UTC+2, Klaus Kragelund wrote:

On Thursday, April 16, 2020 at 12:38:16 AM UTC+2, Clifford Heath wrote:
On 16/4/20 8:24 am, Klaus Kragelund wrote:
Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust

Find an LED that has the reverse temperature dependency and derive bias
from it.

Sounds like a good idea

I could maybe use a dual BJT, and then use the second BJT to set the bias (VBE amplifer style, so the second BJT creates a little higher voltage)

Cheers

Klaus

AOE rev2 figure 2.41 has a circuit like this

 

[Klaus Kragelund]

On Thursday, April 16, 2020 at 12:38:16 AM UTC+2, Clifford Heath wrote:

On 16/4/20 8:24 am, Klaus Kragelund wrote:
Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust

Find an LED that has the reverse temperature dependency and derive bias
from it.

Sounds like a good idea

I could maybe use a dual BJT, and then use the second BJT to set the bias (VBE amplifer style, so the second BJT creates a little higher voltage)

Cheers

Klaus

 

On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
<klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Sounds risky. You could add some gain ahead of the detector for a few
cents more.

This is pleasingly weird:

https://www.dropbox.com/s/akcwlbg744fnds4/Detector_4.jpg?raw=1

(hope the pic works. Dropbox has again chosen to change everything
just to annoy their paying customers.)

Can your digital thing catch moderate duty cycle things? If so,
eliminate Q2.





--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[whit3rd]

On Wednesday, April 15, 2020 at 3:24:24 PM UTC-7, Klaus Kragelund wrote:

... I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

I would like to have it work down at 100mVpeak ... - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias

I've used the collector voltage, through an RC filter, to bias the base.
If you attenuate it, down to circa 0.6V for 2.5V at the collector,
your 5V square wave stabilizes at about 50% duty cycle, and tracks
that duty cycle without much temperature drift.

 

[Jasen Betts]

On 2020-04-15, Klaus Kragelund <klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

you're describing a data slicer.


If the 1Mhz signal is always present
you can just take bias from the collector

-----+-- +5
|
10K
|
--||--+-[100k]---+
| |/
+-[10K]--|
| |e
33K |
| |
+----------+
---
///

somethig like that,

there will be a slight temperature-to-PWM effect at the output and the
100k feeds that back to base bias.

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

--
Jasen.

 

[Klaus Kragelund]

On Thursday, April 16, 2020 at 4:41:28 AM UTC+2, jla...@highlandsniptechnology.com wrote:

On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Sounds risky. You could add some gain ahead of the detector for a few
cents more.

This is pleasingly weird:

https://www.dropbox.com/s/akcwlbg744fnds4/Detector_4.jpg?raw=1

(hope the pic works. Dropbox has again chosen to change everything
just to annoy their paying customers.)

It works fine. (Dropbox has a funny way to annoy their customers)

I am working on your idea, needs some more work. Will post back if it works

Can your digital thing catch moderate duty cycle things? If so,
eliminate Q2.

It's a UART input on a micro, so it needs pretty accurate timing

Cheers

Klaus

 

[Lasse Langwadt Christense]

torsdag den 16. april 2020 kl. 12.42.17 UTC+2 skrev Klaus Kragelund:

On Thursday, April 16, 2020 at 4:41:28 AM UTC+2, jla...@highlandsniptechnology.com wrote:
On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Sounds risky. You could add some gain ahead of the detector for a few
cents more.

This is pleasingly weird:

https://www.dropbox.com/s/akcwlbg744fnds4/Detector_4.jpg?raw=1

(hope the pic works. Dropbox has again chosen to change everything
just to annoy their paying customers.)


It works fine. (Dropbox has a funny way to annoy their customers)

I am working on your idea, needs some more work. Will post back if it works

Can your digital thing catch moderate duty cycle things? If so,
eliminate Q2.


It's a UART input on a micro, so it needs pretty accurate timing

uart has DC is that an issue?

anyway how about a dual transistor and use one of them to set the bias,
sorta like a current mirror?

 

[Winfield Hill]

Klaus Kragelund wrote...

(I am cost driven, so I cannot just use a fast comparator,
which would be the easy solution)

3Peak comparators from LCSC cost as low as 5 cents.


--
Thanks,
- Win

 

[legg]

On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
<klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Normal method is to use a diode or vbe in the biasing
network.

RL

 

On Thu, 16 Apr 2020 03:42:11 -0700 (PDT), Klaus Kragelund
<klauskvik@hotmail.com> wrote:

On Thursday, April 16, 2020 at 4:41:28 AM UTC+2, jla...@highlandsniptechnology.com wrote:
On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Sounds risky. You could add some gain ahead of the detector for a few
cents more.

This is pleasingly weird:

https://www.dropbox.com/s/akcwlbg744fnds4/Detector_4.jpg?raw=1

(hope the pic works. Dropbox has again chosen to change everything
just to annoy their paying customers.)


It works fine. (Dropbox has a funny way to annoy their customers)

I am working on your idea, needs some more work. Will post back if it works

100 mV peak, 200 p-p, should work with healthy margins. Run Q1
saturated with a forced beta around 40 maybe. I've spiced that in the
past and it works.

Can your digital thing catch moderate duty cycle things? If so,
eliminate Q2.


It's a UART input on a micro, so it needs pretty accurate timing

Cheers

Klaus

Oh, you must be modulating serial data on a carrier. I think my
circuit will work for that if the carrier/baud ratio is reasonably
high.




--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Kevin Aylward]

"Klaus Kragelund" wrote in message
news:f3d9b2ae-42a7-429a-93f6-d700addfece4@googlegroups.com...

Hi

So, I have this small detector circuit, where I need to convert a low
voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V
from the collector. The signal is 1MHz, so I have added a baker clamp
schotkky to avoid the BJT >storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to
keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature,
2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to
the actual VBE knee (maybe check the digital side, if no signal, then I
need to shift the bias up, >and if it is clamped to GND, I need to shift
the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be
the easy solution)

I don't believe you. A 100 Mhz ft transistor is a slow transistor, there are
plenty of cheap comparators. As soon as one states adding any components,
such as diodes, it all adds up.

Even making a diff pair would probably be better option. I would research
strandad solutions a bit harder.

-- Kevin Aylward
http://www.anasoft.co.uk - SuperSpice
http://www.kevinaylward.co.uk/ee/index.html

 

[John Larkin]

On Fri, 17 Apr 2020 19:26:39 +0100, "Kevin Aylward"
<kevinRemovAT@kevinaylward.co.uk> wrote:

"Klaus Kragelund" wrote in message
news:f3d9b2ae-42a7-429a-93f6-d700addfece4@googlegroups.com...

Hi

So, I have this small detector circuit, where I need to convert a low
voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V
from the collector. The signal is 1MHz, so I have added a baker clamp
schotkky to avoid the BJT >storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to
keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature,
2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to
the actual VBE knee (maybe check the digital side, if no signal, then I
need to shift the bias up, >and if it is clamped to GND, I need to shift
the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be
the easy solution)

I don't believe you. A 100 Mhz ft transistor is a slow transistor, there are
plenty of cheap comparators. As soon as one states adding any components,
such as diodes, it all adds up.

Even making a diff pair would probably be better option. I would research
strandad solutions a bit harder.

-- Kevin Aylward
http://www.anasoft.co.uk - SuperSpice
http://www.kevinaylward.co.uk/ee/index.html

I think he may be sending async serial data modulated OOK on a 1 MHz
carrier of varying amplitude. ?. That will take more than just a
comparator to translate.

--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Klaus Kragelund]

On Thursday, April 16, 2020 at 4:27:30 PM UTC+2, Winfield Hill wrote:

Klaus Kragelund wrote...

(I am cost driven, so I cannot just use a fast comparator,
which would be the easy solution)

3Peak comparators from LCSC cost as low as 5 cents.

I know. Sadly, no way I can get such a part in a design, purchase would crucify me

 

[Klaus Kragelund]

On Thursday, April 16, 2020 at 5:56:42 PM UTC+2, jla...@highlandsniptechnology.com wrote:

On Thu, 16 Apr 2020 03:42:11 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

On Thursday, April 16, 2020 at 4:41:28 AM UTC+2, jla...@highlandsniptechnology.com wrote:
On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Sounds risky. You could add some gain ahead of the detector for a few
cents more.

This is pleasingly weird:

https://www.dropbox.com/s/akcwlbg744fnds4/Detector_4.jpg?raw=1

(hope the pic works. Dropbox has again chosen to change everything
just to annoy their paying customers.)


It works fine. (Dropbox has a funny way to annoy their customers)

I am working on your idea, needs some more work. Will post back if it works

100 mV peak, 200 p-p, should work with healthy margins. Run Q1
saturated with a forced beta around 40 maybe. I've spiced that in the
past and it works.

I used forced beta at 30, and tried different values, but could not get it to work

Can your digital thing catch moderate duty cycle things? If so,
eliminate Q2.


It's a UART input on a micro, so it needs pretty accurate timing

Cheers

Klaus

Oh, you must be modulating serial data on a carrier. I think my
circuit will work for that if the carrier/baud ratio is reasonably
high.

Yes, correct. I am modulating a carrier with a UART TX signal. Then trying to pick it up at the other end with a sort of a AM modulation scheme. I have a cap on the output of the last BJT emitter, with a pull-up, so that I have a hold function when the NPN is turned on a regular intervals

Cheers

Klaus

 

[Klaus Kragelund]

On Friday, April 17, 2020 at 8:26:44 PM UTC+2, Kevin Aylward wrote:

"Klaus Kragelund" wrote in message
news:f3d9b2ae-42a7-429a-93f6-d700addfece4@googlegroups.com...

Hi

So, I have this small detector circuit, where I need to convert a low
voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V
from the collector. The signal is 1MHz, so I have added a baker clamp
schotkky to avoid the BJT >storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to
keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature,
2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to
the actual VBE knee (maybe check the digital side, if no signal, then I
need to shift the bias up, >and if it is clamped to GND, I need to shift
the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be
the easy solution)

I don't believe you. A 100 Mhz ft transistor is a slow transistor, there are
plenty of cheap comparators. As soon as one states adding any components,
such as diodes, it all adds up.

NPN or diodes are much below 0.01 EUR. A comparator with less than 100ns delay is close to 0.1 EUR

Even making a diff pair would probably be better option. I would research
strandad solutions a bit harder.

A diff pair could be a good idea to try out. If I can keep one transistor out of deep saturation

Cheers

Klaus

 

[Klaus Kragelund]

On Friday, April 17, 2020 at 9:26:42 PM UTC+2, John Larkin wrote:

On Fri, 17 Apr 2020 19:26:39 +0100, "Kevin Aylward"
kevinRemovAT@kevinaylward.co.uk> wrote:

"Klaus Kragelund" wrote in message
news:f3d9b2ae-42a7-429a-93f6-d700addfece4@googlegroups.com...

Hi

So, I have this small detector circuit, where I need to convert a low
voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V
from the collector. The signal is 1MHz, so I have added a baker clamp
schotkky to avoid the BJT >storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to
keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature,
2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to
the actual VBE knee (maybe check the digital side, if no signal, then I
need to shift the bias up, >and if it is clamped to GND, I need to shift
the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be
the easy solution)

I don't believe you. A 100 Mhz ft transistor is a slow transistor, there are
plenty of cheap comparators. As soon as one states adding any components,
such as diodes, it all adds up.

Even making a diff pair would probably be better option. I would research
strandad solutions a bit harder.

-- Kevin Aylward
http://www.anasoft.co.uk - SuperSpice
http://www.kevinaylward.co.uk/ee/index.html


I think he may be sending async serial data modulated OOK on a 1 MHz
carrier of varying amplitude. ?. That will take more than just a
comparator to translate.

Yes, OOK it is

It can be done with a comparator. Just need a hold capacitor on the output and a schmitt trigger input on the UART

 

[John Larkin]

On Sat, 18 Apr 2020 10:19:55 -0700 (PDT), Klaus Kragelund
<klauskvik@hotmail.com> wrote:

On Friday, April 17, 2020 at 8:26:44 PM UTC+2, Kevin Aylward wrote:
"Klaus Kragelund" wrote in message
news:f3d9b2ae-42a7-429a-93f6-d700addfece4@googlegroups.com...

Hi

So, I have this small detector circuit, where I need to convert a low
voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V
from the collector. The signal is 1MHz, so I have added a baker clamp
schotkky to avoid the BJT >storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to
keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature,
2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to
the actual VBE knee (maybe check the digital side, if no signal, then I
need to shift the bias up, >and if it is clamped to GND, I need to shift
the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be
the easy solution)

I don't believe you. A 100 Mhz ft transistor is a slow transistor, there are
plenty of cheap comparators. As soon as one states adding any components,
such as diodes, it all adds up.


NPN or diodes are much below 0.01 EUR. A comparator with less than 100ns delay is close to 0.1 EUR

Even making a diff pair would probably be better option. I would research
strandad solutions a bit harder.

An LVDS line receiver, or an opamp, can be a good comparator. I think
there are cheap opamps.

I like the zoo sound. OOK! OOK!


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Tauno Voipio]

On 18.4.20 20:17, Klaus Kragelund wrote:

On Thursday, April 16, 2020 at 5:56:42 PM UTC+2, jla...@highlandsniptechnology.com wrote:
On Thu, 16 Apr 2020 03:42:11 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

On Thursday, April 16, 2020 at 4:41:28 AM UTC+2, jla...@highlandsniptechnology.com wrote:
On Wed, 15 Apr 2020 15:24:19 -0700 (PDT), Klaus Kragelund
klauskvik@hotmail.com> wrote:

Hi

So, I have this small detector circuit, where I need to convert a low voltage (1Vpeak) sine wave into a digital signal

I do that by feeding it into the base of a BC847, with a resistor to 5V from the collector. The signal is 1MHz, so I have added a baker clamp schotkky to avoid the BJT storage time

It works fine

I would like to have it work down at 100mVpeak instead

So I capacitive couple the signal and add a bias network at the base to keep it biased just below the point where the BJT knee kicks in (600mV)

That also works fine - but the VBE voltage shifts a lot with temperature, 2mV/K

So I was thinking some kind of auto adjust to adjust the bias according to the actual VBE knee (maybe check the digital side, if no signal, then I need to shift the bias up, and if it is clamped to GND, I need to shift the bias down)

Anyone done something like this?

(I am cost driven, so I cannot just use a fast comparator, which would be the easy solution)

Cheers

Klaus

Sounds risky. You could add some gain ahead of the detector for a few
cents more.

This is pleasingly weird:

https://www.dropbox.com/s/akcwlbg744fnds4/Detector_4.jpg?raw=1

(hope the pic works. Dropbox has again chosen to change everything
just to annoy their paying customers.)


It works fine. (Dropbox has a funny way to annoy their customers)

I am working on your idea, needs some more work. Will post back if it works

100 mV peak, 200 p-p, should work with healthy margins. Run Q1
saturated with a forced beta around 40 maybe. I've spiced that in the
past and it works.


I used forced beta at 30, and tried different values, but could not get it to work


Can your digital thing catch moderate duty cycle things? If so,
eliminate Q2.


It's a UART input on a micro, so it needs pretty accurate timing

Cheers

Klaus

Oh, you must be modulating serial data on a carrier. I think my
circuit will work for that if the carrier/baud ratio is reasonably
high.


Yes, correct. I am modulating a carrier with a UART TX signal. Then trying to pick it up at the other end with a sort of a AM modulation scheme. I have a cap on the output of the last BJT emitter, with a pull-up, so that I have a hold function when the NPN is turned on a regular intervals

Cheers

Klaus

AM is not a good idea for digital transmission. It was understood
in radio transmissions decades ago and moved to FSK.

It is easy to detect the onset of the AM pulse, but the absence
of the pulse (or low level) is difficult to relaibly detect.

--

-TV