replacing a 1:1 output x-former with a that 1646 chip

[pucho812]

o.k. trying to figure this one out. got a preamp that is a 1:8 input transformer feeding a 5534 preamp, which feeds the 1:1(600Ω) output transformer to balance out. When I measure the circuit I can get about 63db of gain of 1K  at the output of the transformer. When I  replaced the 1:! transformer and wire in the 1646 per datasheet,  it works but I can a max output of 50dB and the 1646 gets really hot.  I would assume the 5534 could drive the 1646 easily.  I am sure there is something I am missing and know it's not only cj's scorn for wanting to get rid of the transformer in place of an inexpensive chip.

 

[JohnRoberts]

chip hot bad... figure out why...

JR

 

[pucho812]

hmmm that data sheet.

http://www.thatcorp.com/datashts/THAT_1606-1646_Datasheet.pdf

and the schematic of the preamp in question, the op-amp is a 5534 not a 2520 as per the schematic

The input transformer is 1:8 and the output transformer is 1:1 600Ω:600Ω which gives me a max of 63 db preamp gain

but switching to the that I get a a max of 50db of gain and the hot chip... I know I am missing something simple but I would think the 5534 could drive the load of the 1646...

 

[mjrippe]

Odd, you should have 5-6dB *more* gain with the 1646.  What rail voltages are you using and what load is on the output?  Double check your pinout and wiring on the 1646.

 

[usekgb]

Interesting.  I've done something very similar using a drv134, and it worked just fine.  Is you THAT chip oscillating?  Is there DC on the input?  Did you use power supply decoupling caps?  All things to check.

 

[usekgb]

What kind of load are you trying to drive?

 

[pucho812]

After the chip was the ap tester at work. It was wired per data sheet although no coupling caps on the rails were installed.  The chip was too hot to touch for any length of time. the power rails were +/-18vdc so it should have been fine  I have the schematic above I must be missing somethin or is it simply no coupling caps?

 

[SSLtech]

"Heat"...

Don't think volts.

Think Amps.

 

[pucho812]

"Heat"...

Don't think volts.

Think Amps.

right and boy was it hot.  are you suggesting the 5534 was pushing to much amps?

 

[JohnRoberts]

The chip getting hot probably has some kind of load fault connected to it's output... trace the output connections first.

I think some of those output chops can oscillate if not connected per data sheet advice.

JR

 

[PRR]

My 2 cents is on wrong power polarity.

Almost nothing else would make that chip HOT.

 

[bruce0]

Lots of smart guys on this thread, and they are probably right about the heat issue being the primary issue.
So yeah check for DC, check for polarity, etc.

But I wonder....
A 1646 replaces a 1:2 transformer not a 1:1 transformer because the 1646 adds 6db ( or 5.3dB into 600 ohms)
Furthermore, a transformer sees an unbalanced load and a balanced load and treats them identically.  A 1646 does not.
Furthermore a transformer treats a hot signal on the primary "gracefully" (you can make it distort but transformers don't clip) but a 1646 will half the output if you drive unbalanced.
So are you driving a oscilloscope with one side connected to a grounded ground clip?

So what are you driving with this thing (during your test).  Is it driving unbalanced or balanced? what is the load?

I suggest you tell us a bit about your test environment and the actual test values.  Rather than expressed in terms of gain, expressed in terms of RMS Volts or dBu of output.  Tell us what load you are driving. The power rails would limit the output dBu to an unbalanced output (one side shorted) to about 25 volts RMS (30dBu).  The same test into balanced would double the output.

from the 1646 datasheet.

Gain structure
The 1606 and 1646 both provide +6 dB gain
(factor of 2) between their inputs and differential
outputs. This is appropriate, since with a balanced
output, twice the voltage between the power supply
rails is available at the output of the stage. The
single-ended input of the 1646 can accept signals
that swing to nearly the power supply rails without
distortion, when driving into a differential (floating)
load. The balanced input of the 1606 can accept
signals at each input that swing to nearly one-half the
power supply rails without distortion, when driving
into differential loads.
Both devices, when driving single-ended loads,
will clip at about half the output voltage as compared
to a differential load. This is because only one of the
two output signals will be available. Despite the
output clipping, the input to the devices does not
need to be constrained - they will work without
undue problems being overdriven at their inputs
when the outputs are clipping into single-ended
loads.

 

[bruce0]

My 2 cents is on wrong power polarity.

Almost nothing else would make that chip HOT.

I would be surprised if it could do 50dB gain with power rails inverted.

volker maybe you are right, but he has 50db of gain....but maybe thqt is 63 in the 5534 being droped by the chip attenutor.

 

[volker]

My 2 cents is on wrong power polarity.

Almost nothing else would make that chip HOT.

I would be surprised if it could do 50dB gain with power rails inverted.

With the semiconductor part of the IC doing more or less nothing with inverted rails, the rest of the chip acts like a passive attenuator. 63dB down to 50dB would be in the ballpark judging from the internal circuit diagram.

 

[pucho812]

The reason I am using the 1646 is the fact we have them in mass quantities  in the shop.

o.k. thanks for the suggestions guys, I'll be able to get back on this in a few days.  But just so I am 100% clear

Per the data sheet
Pin 1 out -
pin 2 sens -
Pin 3 is gnd
pin 4 is input
pin 5 vee(V-)
pin 6 Vcc(V+)
pin 7 Sns+
pin 8 Out +

I am thinking I wired my off board chip incorrectly and need to recheck those connections.

so I should  wire  the 1646 as follows

Tie pins 1 & 2 together  so that they end up at pin 3 of the XLR
Pins 7 & 8  tie together so that they end up at pin 2 of the XLR

Pin 4 of the that 1646 would connect right  where the primary of the output transformer goes

Pin 5 would connect to the V- rail
pin 6 would connect to the v+ rail

and pin 3 would connect to ground....

correct? I should be o.k. using the chip in the above application.

 

[JohnRoberts]

While I do not have much experience with those canned I/O parts Wayne Kirkwood has done some app work with THAT and mentioned that the 1646 is sensitive to series inductance (like a long wire) in pin 3.

Follow the app note advice closely, maybe beef up pin 3 ground connection.

JR

Figure 4 shows the most basic connection for a 1646. The only external components needed are the local 100 nF bypass capacitors. These should be within 1 inch of the 1646 pins.

 

[pucho812]

Update,

After some checking. I had the unit wired wrong... Silly me. 
I now get the voltage in crease.

From the ap machine

So for example if I am feeding it a .775vrms signa @ 1k sine wave I measure 2.1 volts  at the output. However  I am noticing a lot of white noise/hiss in the background on the scope which is audible if I plug it into the line in on thre test mixer we use a mackie 1202. ... hmmmmm

 

[volker]

So for example if I am feeding it a .775vrms signal I measure 2.1 volts  at the output. However  I am noticing a lot of white noise/hiss in the background... hmmmmm

Did you try a new part, not the miswired one?

 

[pucho812]

Yes.

 

[pucho812]

I must be doing something really stupid. If I replace the that 1646 with a drV 134 in the same setup. No noise.  This is with brand new ic's.... hmmmmm

might  just do the 134 then....