Sara92
Newbie level 5
- Joined
- Jan 5, 2011
- Messages
- 8
- Helped
- 0
- Reputation
- 0
- Reaction score
- 0
- Trophy points
- 1,281
- Activity points
- 1,358
Hi!
Sorry for bothering you again. A few hours ago I posted a thread regarding the use of an LVDS Receiver as comparator. Well, you told me why it's not a good idea.
Anyway, it might be more helpful, if I described my problem here:
I need to sample a differential signal and can't use an AD converter. So it's basically about 1bit sampling.
The inout is a rather noisy background with occasional (triggered) signal bursts. They have a frequency of about 100MHz and are about 100ns long. Sampling frequency is 1.5GHz.
With some
averaging this works rather well. The noise background is detected as 1 or 0 each with 50% so the averaged sample has the value n/2. Positive signals always get a 1 so they end up with value n while negative values sum up to zero. Unfortunately I see sometimes bursts which I would like to suppress. They have a much smaller amplitude (factor 10) than the target signals but they are still not burried in the noise.
Is there anyway to do 1Bit sampling with suppression of noise or unwanted signals?
I considered using two 1Bit inputs. Instead of having zero as threshold I would like to use threshold +-100mV. That means if the signal is >100mV I would like to add 1 to my sample, if it's below -100mV it would add -1 and if it's in between zero is added. So noise and small bursts would be a flat line=0, the positive signals would sum up as n and the negatives as -n.
I thought of using comparators for this porpuse but at these fequencies (1.5GHz) they are expensive and power hungry. (That's where the Idea with LVDS receivers as comparator came from). :grin:
A limter amplifier with a "blind" input range from -100mV to +100mV would also be nice. I could use this to detect large positives and negatives spikes.
Something like limiter amplifier with a fast loss-of-signal output might also help. I could use this output to decide if i see a "real" burst. But the LOS outputs tend to be very slow..
Has anybody got an idea?
:roll:
Sorry for bothering you again. A few hours ago I posted a thread regarding the use of an LVDS Receiver as comparator. Well, you told me why it's not a good idea.
Anyway, it might be more helpful, if I described my problem here:
I need to sample a differential signal and can't use an AD converter. So it's basically about 1bit sampling.
The inout is a rather noisy background with occasional (triggered) signal bursts. They have a frequency of about 100MHz and are about 100ns long. Sampling frequency is 1.5GHz.
With some
averaging this works rather well. The noise background is detected as 1 or 0 each with 50% so the averaged sample has the value n/2. Positive signals always get a 1 so they end up with value n while negative values sum up to zero. Unfortunately I see sometimes bursts which I would like to suppress. They have a much smaller amplitude (factor 10) than the target signals but they are still not burried in the noise.Is there anyway to do 1Bit sampling with suppression of noise or unwanted signals?
I considered using two 1Bit inputs. Instead of having zero as threshold I would like to use threshold +-100mV. That means if the signal is >100mV I would like to add 1 to my sample, if it's below -100mV it would add -1 and if it's in between zero is added. So noise and small bursts would be a flat line=0, the positive signals would sum up as n and the negatives as -n.
I thought of using comparators for this porpuse but at these fequencies (1.5GHz) they are expensive and power hungry. (That's where the Idea with LVDS receivers as comparator came from). :grin:
A limter amplifier with a "blind" input range from -100mV to +100mV would also be nice. I could use this to detect large positives and negatives spikes.
Something like limiter amplifier with a fast loss-of-signal output might also help. I could use this output to decide if i see a "real" burst. But the LOS outputs tend to be very slow..
Has anybody got an idea?
:roll:
