[hpsdr] ALEX - Input protection

Tue Aug 7 17:42:10 PDT 2007

Hi Tommy, All,

Tommy, I do have the same problem, keeping track of who said what and
where..... ;-)
On the subject now.....

As far as I understand it, I think that what you would be aiming for would
ideally be a device with

- zero response time (0 nsec) and 
- infinite current sinking capability
- a clamping voltage equal to the breakdown voltage regardless of the
current being sinked

when the breakdown voltage is reached.

This is of course not realistic in a real world.
Not sure but I think the capacitance slows down the response time (the
EMI-pulse would have to charge the capacitor first up to the breakdown
voltage?) and you would like it therefore to be as low as possible?
Or would you like to have it as low as possible to have it not shunt the RF
input too much (Low C = High Xc)? In any way, it shouldn't be "visible" in
normal operation, as if it weren't there.

The limited current sinking capability of course causes the device to blow
up if a too high current passes through the device when the breakdown
voltage is reached.
And I think an increased current sinking capability goes with a higher
capacitance value?

So I think you have to make a compromise in selecting a device; required
response time or capacitance versus the required amount of current it can
sink, depending on your particular situation.
This is I think also the reason why in the receiver that Frank describes you
can choose for different numbers of diodes/transzorbs configurations?

But paralleling devices gives you besides higher current sinking also higher
capacitance values.
Anti-paralleling (one diode reverse and parallel to the other diode) gives
you protection for positive and negative pulses and I think would double
capacitance.
Putting them in series will give a lower capacitance but a higher breakdown
and clamping voltage I think.

I agree with you that another practical consideration could be the
availability and pricing of the parts.

I know these are not hard answers to your questions, but I hope it maybe
helps a little in understanding the matter.
Others please comment also, as I'm not really familiar with the technology
involved in this field.

http://people.seas.harvard.edu/~jones/es154/lectures/lecture_2/breakdown/bre
akdown.html

I'm learning also....again! ;-)

73 's,

Henry.

> -----Original Message-----
> From: hpsdr-bounces at lists.hpsdr.org 
> [mailto:hpsdr-bounces at lists.hpsdr.org] On Behalf Of 
> Hollingsworth, Tommy
> Sent: dinsdag 7 augustus 2007 18:42
> To: Graham Haddock
> Cc: hpsdr at hpsdr.org
> Subject: Re: [hpsdr] ALEX - Input protection
> 
> 
> ***** High Performance Software Defined Radio Discussion List *****
> 
> Graham:
> I am trying to understand what your criteria are.  I just 
> completed an SDR
> 1000 clone and am trying to decide what kind of clamp to use 
> on the input.
> After spending 6 months building it, I really do not want the 
> smoke to get
> out of my working radio :)  Protek parts are not easy for me 
> to obtain.
> The parts below look interesting:
> 
> 		Semtech	Semtech	DiodesInc	Protek
> 		RClamp0502B	SL05		DLP05LC		GBLC05C 
> V standoff	 5		2.8		5		 5
> volts
> Vbr		 6		6		6		 6
> volts
> Vclamp	15		5.5		11		
> 18.3	volts
> I peak	 5		24		17		
> 17 	amps
> C		.6		2		1.6		
>  3 	pf
> 
> Package	SC75		SOIC8		SOT23		SOD323
> 
>    ***Note that Protek uses typical so I used typical for all C values
> Note that similar parts are also available from On 
> Semiconductor.  I really
> doubt that these are more expensive than the Protek part, considering
> Protek's $50 minimum per part and $500 minimum order. Note 
> that all of these
> parts are available from Digikey except for Protek. I assume 
> next that you
> have a target current/energy level higher than 5 amps so I 
> looked for some
> in that range.  The Diodes Inc DLP05LC looks really good to 
> me at 17 amps.
> You could have two in parallel for the capacitance of the 
> GBL05C .  Note
> that I see it as a good design idea to create a bidirectional 
> clamp from two
> unidirectional devices.  The spike really does ring, and in 
> doing so dumps
> energy into both device, not symmetrically but close enough. 
> 
> I am a little concerned that by trying to push the 
> capacitance to a really
> low value, a TVS is being chosen that will not really be 
> robust enough.
> Henry Vredegoor described using one or multiple 1N4148 diodes 
> in series with
> large TVS parts.  Note that the 1N4148 has a capacitance of 
> 4.0 pF(max) and
> the 1N4448 has a capacitance of 2 pF.  I am curious if we can 
> stack two of
> the TVS in series and cut the relative capacitance in half.  I am not
> certain that it would work that way.  I do have a transient 
> tester available
> and may have to build a test board to find out. 
> 
> This sounds like we would be just as well off putting several 
> of the DLP05LC
> TVS parts in parallel on the line and forgetting about the 
> smaller diodes.
> I am really curious what Henry and some of the others might 
> feel about this
> part, considering it is easy to obtain.  Is this clamp 
> voltage high enough?
> 
> Please let me know if this makes sense, or if I am 
> overlooking something.  I
> am anxious to get my radio on the air, but I plan on it 
> lasting for a long
> time so I won't rush it.  I also plan on building the Mercury 
> and Penelope
> in the near future.  
> Cheers
> 73
> Tommy Hollingsworth KV5T

-------------- next part --------------
A non-text attachment was scrubbed...
Name: logo[1].jpg
Type: image/jpeg
Size: 41309 bytes
Desc: not available
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20070808/7bb30a09/attachment-0004.jpg>