The other day, while cleaning out some very old reports, I came across the mention of a piece of test gear I hadn’t thought about in ages: the Gertsch Ratio Transformer. Way, way back in the day, I was involved with interface circuitry for LVDTs, and this basic passive box was a key part of the test-and-validation process.
If you’re not familiar with the LVDT (linear variable differential transformer) it’s a very rugged, reliable, and precise position transducer that can be used over a wide measurement range — spans of 6 or even 10 inches are common. It’s completely passive. You excite the primary coil with an AC signal, and read back the secondary-side signals, then do some fairly straightforward demodulation to get a signal proportional to position.
To check out the accuracy, repeatability, and precision of our interface circuitry, we initially had a long-stroke LVDT set up in a fixture with a leadscrew on the movable core, along with a precision caliper for visual readout. We’d manually set the position at various settings via the leadscrew, and then check our electronics compared to the caliper reading. It worked, but it required a lot of manual setting, along with a need to pay attention when setting the position and reading the caliper.
Then one of the old-timers said, “Hey, get a ratio transformer; it’ll make your life easier.” A few hundred dollars later (the boss signed off, thankfully) and we were all set. The Gertsch Ratio Transformer is just that: a precision, multitap autotransformer box with big knobs to set the input/output ratio. To our electronics, it looked just like the LVDT, except it was much easier to consistently set to different values compared to adjusting the real LVDT.
with big knobs to set the input/output ratio.
Spurred by remembering the ratio transformer cited in that old report, I used the Internet to step into the “way-back” machine. Sure enough, there were several good links, including a site devoted to them here, some brand acquisition history here, as well as some units for sale (along with tutorial) on eBay here.
Seeing them for sale reiterated one point about them: Unless they have been electrically abused or left outside, there is nothing to go wrong with them. Even an old one that has been used repeatedly should still work, and the same is true for one that has been sitting in a storeroom and forgotten. No batteries or capacitors to leak or dry out, no operating system to go obsolete, no LCDs to decompose, just a combination of wire and switches. Perhaps the switch contacts have some modest corrosion that can be cleaned, but that's all there would be to worry about.
There’s a lesson here. Sometimes, simulating your real-world signal source or transducer is key to evaluating the performance of your analog front end, in design, debug, and qualification stages. That's why a modest investment in a thermocouple simulator, for example, is a great idea: You don’t want to be messing around and heating your thermocouples, for obvious reasons. But simulators for other real-work transducers and signal patterns — pressure, position, velocity, loudness; the list goes on — can really be worth their modest cost. Of course, some are even more sophisticated, as you can also get standardized, certified simulators for an entire array of cardiac signals and their permutations, very helpful for validating your ECG device or heart-rate monitor.
There’s another thing about that ratio transformer that I recall: the nice, solid, tangible click of its front-panel switches as you stepped through the various settings. It really made you aware that you were dealing with signals that live in the real world, even if you were only simulating them. Often, these signals are quickly digitized and then consumed within the system software, to be just another bunch of anonymous bits. Yet you always need to have some way to remind you that you are working with transducer and analog I/O signals, which represent motion, power, action, and perhaps even danger.
Have you ever used a basic yet effective simulator or some sort of hardware-in-the-loop (HITL) to move your project along? Did you ever have to build your own when you couldn’t buy what you needed?
Hi Bill–not sure if you would put this in the simulator category, but way back in my past I was setting up a measurement system which included some 4-20 mA output transducers (mostly pressure). Not having direct interface modules for current measurement, we always used precision resistors to create a voltage, and measured the voltage with A/D converters.
During development of the software, we needed to provide inputs to the A/D. For example, we had to write code for the calibration and measurement phases. Solution? A precision current source. That plus a resistor could be used to simulate any process transducer providing 4-20 mA (which was an is a lot of them).
Bill
Almost every project I work with needs some form of simulation. Often it is just a series of switches or nothing more than a potentiometer. When I need accurate analog signals the best tool I have is a Fluke 741B Documenting Process Calibrator, along with my 'scope as my favourite tool on my bench. Will simulate, RTDS/thermocouples, 4-20mA, AC waveforms, resistances, voltages and more, not to mention measure them as well (simultaneously)!
Probably the most complex simulator I ever built was for a gas furnace. We had to simulate the blowers and the igniters and the signals the the igniter saw that allowed it to operate or detect a failure. It included a real blower to get the timing right for the air pressure switch. And there were 6 of these in parallel. It was real noisy and quite windy in the immediate vicinity.
One of the more complex things to simulate is elapsed time. Some of the products that we make have processes that take hours, and in the case of service records months or even years. I normally do this through switches in software to accelerate the counters, but even then time timescales may need several different settings.
A board that I redesigned a few years ago required a syncho-resolver simulator so that the new board can be confirmed as reading proper angles of position. Nice thing was that the board was ±1 bit for a 12-bit system.
@Bill–Your nice article sparked another line of thought. Not directly realted to analog, but certainly not “digital”, is the issue of simulating the environment in which the end product will be placed. Millions of dollars are spent every year paying for environmental chambers, thermal shock chambers, shock and vibe equipment, EMI/ESD test devices, and so on. In addition, more millions are spent on paying test labs to run certification tests, such as temp-humidity validation.
In one startup of which I was the head of engineering, we didn't have much of a budget (a bit different from many startups today, that attract zillions of dollars on just an idea…). We were making electronic components that would be used in various enviornments from indoors to on cars to exposure to salt spray from the ocean.
I recall two particular environmental simulators we devised. In one, we purchased chicken egg incubators; the cheapest ones we could get; they were styrofoam boxes with a heater, a simple controller, analog temperature gauge, and a small window to peek in and see if anything had hatched. We put salt water in open cups in them, cranked up the temperature, and proceeded to cycle parts in and. We knew the simulation was working becuase we corroded and killed the heaters quite well.
The other high-tech simulation was mounting an antenna which in use would be on top of a vehicle, onto the top of a large plastic storage bin, the kind with the lid that snaps on and is relatively sealed. We mounted the bottom half in the pickup bed of an employee, holding it in place with a few 25 lb dumbells. This allowed us to drive the antenna through car washes then snap the lid off and take it back in the lab for evaluation. We ran units like this through a car wash every day for a couple of months. The employee had a perpetually clean car, and we had a modicum of confidence in our design.
I think that a good compromise between the costs of simulation tools and the budget is the key to successful design in the real business environment.
“One of the more complex things to simulate is elapsed time. Some of the products that we make have processes that take hours, and in the case of service records months or even years.”
@antedelivian: I've faced that problem too particularly while designing simulations for software. At best you can use the CPU cycles as a proxy for time but that's a highly inaccurate measure in many cases.
@eafpres1: I think the combination of hardware and software make it seem like a really complex simulation. But it must have its own advantages over electronic simulations that are purely software based. I only have experience of working with the latter.
“In one, we purchased chicken egg incubators; the cheapest ones we could get;”
@Eafpres1: That sounds like a really ingenious and a cheap way of simulating a high temperature environment. I guess this can be utilized at a number of places where companies have to test how their products will behave in natural heat.
>> hardware-in-the-loop (HITL)
First, thank you for teaching me about this. Not aware that HITL is hardware-in-the-loop. Over the years, I have come to agree that Cadence for all its costs in circuit design has few competitors when you need serious work. You can get along with LTSpice and other freewares but when complexity sets in, you can be wasting time.
During initial development period time, simulation has benefit for capturing all bug and analog design that engineer had missed. And then, once the system is passed in the simulation environment, in real environment, the product will be reevaluated. For example, PCB design and simulation could be done on Desktop to reduce design complexity by using simulation data. Then, PCB board will be tested under real data interacting with real environment.
I have used coolspice, recently they have included SOA i.e.safe operating area. Any idea how exactly it works.
Among so many engineering tools, Cadence PCB design tool also provides auto route and generates a variety manufacture format with etch editing. When comparing it with other PCB design tool, there are many option for component layer.
@DaeJ; Yes true there are many components but there are some other factors too to be looked into before going further.
1. Cost factor
2. Compatibility factor
3. Reliability factor
4. Cross funcional facor
etnapowers true if it cant afford to the users what is the point of designing.
However tzubair combination of hardware and Software together will increase the accuracy.
chirshadblog , I also do agree with you on this. If it going to be use in a business manner cost factor is very important.
@tzubair–I think it depends a lot on what you are simulating and for what reason. But the more I think about it, for the purposes of software development we could have written a module to provide data to the measurement subroutine. I'm involved a project now and we routinely do this–until some other parts of the software are ready, or until we get the remote sensing hardware, we write emulators/simulators to feed data to a listener in the application.
In my experience when first involved in antenna design, in the 1990's, there were lots of antenna design guys who had developed approaches or instincts about certain problems. This worked OK when antennas were simple and large. But as the mobile wireless movement took over, we had to put antennas into things that were way too small, and in other compromising environments. For many years in cell phones antennas were design by trial and error on the bench with wire, copper tape, etc. The first attempts to simulate antenna desings on mobile devices failed becuase the old guard defined the problem as if there was a large groundplane, no external loading to change impedance, etc.
Eventually HFSS and other tools became mreo common in antennas for mobile and a lot of old designers found out that in small devices the radiating currents were actually on the device–the “antenna” was really an impedance matching lumped element that coupled the RF output to free space allowing currents to run all over the circuit boards. Finally pepole learned from simulations showed why some phones were really sensitive to hands or holding near head, and engineers started to figure out how to control where the radiating currents were and design better RF perforamnce.
In my view without simulation we would still be sitting at the bench with a roll of copper tape and Xacto knife.
“I'm involved a project now and we routinely do this–until some other parts of the software are ready, or until we get the remote sensing hardware, we write emulators/simulators to feed data to a listener in the application.”
@eafpres1: Is the input to the application a voice-based command? I've done that once in my a career. We had a project where we had to program a small robot to act on human voice commands. We simulated the commands from a computer because it required the ability to give commands in different accents and none of us could afford to gather as many human beings with such different voice accents.
@tzubair: Yes indeed that is the issue with voice recognition. Not sure whether it will ever have a solution since its practically impossible to decode the voice.
@Ranasinghe: Yes indeed cost is a factor but you do need to consider the roi factor as well.
@tzubair–my current project is not voice control, but there are remote devices, and a system that makes IP connections to them and downloads raw data. Those data are captured by a process that makes the data available to a higher level applicaiton via an API. Depending on what we are working on, the raw data may not be available as we don't have the hardware in place. In that case we simulate a node providing raw data. In other cases, the API may not be ready, but we have the definition of what the requests are and what is returned, so we simulate the API publishing data so the developers can code the higher level application. All these steps allow us to work somewhat in parallel instead of waiting to build the system from the bottom up.
For sure the users of the simulator have to work with a simulation tool good enough to realize effectively the project.
I add that a good systematic approach may enhance the accuracy.
@tzubair: interesting project, I guess you enjoyed working on this item. I think that an accurate system of voice recognition may be utilized in safety application, due to the wealth of voice accents that you described
Even though I will probably never get the chance to use this particular ratio transformer, I also find some very useful information that is relevant to the modern day designer. The key product from the creation of this ratio transformer, even though it may not have been the express objective of its designers, is a device that can live for very long without failing in its precision or being rendered obsolete. Just like the slide rule, even though better gadgets may replace it, it will still remain just as capable of doing its work today as it was 2 decades ago.
It is quite interesting how the idea of simulating works and how it can be used to evaluate the performance of analog front end, debug and even in design. I also like the idea that it help to prevent the risk of heating of thermocouples. But the most future that impress me more is that, Simulators are also very helpful for validating ECG device or heart – rate – monitor.
Calibration and simulation should not be viewed as competing each other but rather as processes that support each other. With every new design, the first process should be the simulation of the design to see how it would actually perform. By changing the parameters within the simulation you can establish the optimal measurements and then use these when you get to the calibration stage. Systematic combination and use of these processes will result in more efficient designs.
Cadence for all its costs in circuit design has few competitors when you need serious work. You can get along with LTSpice and other freewares but when complexity sets in, you can be wasting time.
@goafrit2, I totally agree with you on this. Cadence is very costly tool but then the features they provide are very useful and such features are hard to implement using free tools like LTSpice.
I have used coolspice, recently they have included SOA i.e.safe operating area.
@samicksha, thanks for introducing me to this new tool. How different is this tool from LTSpice ?
I think that a good compromise between the costs of simulation tools and the budget is the key to successful design in the real business environment.
@etnapowers, I totally agree with you. But sometimes we have to compromise on cost because some of the tools available in semiconductor industry are very costly compared to other software tools. For example cadence tools are very costly but its worth buying them instead of using freewares.
I think the combination of hardware and software make it seem like a really complex simulation.
@tzubair, true but hardware and software combination is more prone to errors than just standalone software solution.
LTspice was originally called SwitcherCAD and is sometimes still called by that name.It is node-unlimited and 3rd party models can be imported. Circuit simulations based on transient, AC, noise and DC analysis can be plotted as well as Fourier analysis. Heat dissipation of components can be calculated and efficiency reports can also be generated.
Yes, you're correct, it's the reason for why there's the need of a good balancing between the cost of the tool and the possible revenues coming from the product, which will be sold once designed.
The cost of the tool is not always corresponding to the accuracy of the result. A good engineer has to choose the tool taking into account the budget available and of the level of accuracy required for the product under design.
Cadence is very costly tool but then the features they provide are very useful and such features are hard to implement using free tools like LTSpice.
LTSPICE gets into trouble when you have about 50k transistors. I think it is a great tool but the level of sophistication we need in real practical design environment can only be provided by Cadence.
For example cadence tools are very costly but its worth buying them instead of using freewares.
If you can afford Cadence, it means you are certainly not a startup but generating revenue. In efficiency and performance, it wins. But when it is to show a prototype for that first money, Tanner EDA and other cheaper tools can come handy.
. A good engineer has to choose the tool taking into account the budget available and of the level of accuracy required for the product under design.
But some tools are so good that that argument loses strength. If you have used Cadence for Monte Carlo simulation and try to do same in any other CAD, you will agree that investing in CAD could be the begining of innovation in any firm. Great CAD can unlock creativity beecause it makes your designers understand their works better.
@fasmicro you're right about Candence, Candence's PSpice simulation technology is an advanced technology and industry-proven. Its capable of simulating design from power supplies to simpler ICs, it enables an efficient simulation process and simplifies the viewing of simulation results- analog and digital.
@fasmicro >> Great CAD can unlock creativity beecause it makes your designers understand their works better .
These tools also bring some light when looking for failure causes under stimulus which are difficult to reproduce in practice or when these stimulus could lead to expensive hardware destruction.
Unfortunately many engineers use CAD tools too to tweak some values until they obtain some specific desired response. It is relatively fast to do and reduces the need to have deep knowledge about analog and digital circuit design.
I just saw this auto-cranking simulator from TI, should be pretty handy:
http://www.ti.com/tool/pmp7233?DCMP=tidesigns-gma-auto-pmp7233-en&HQS=tidesigns-gma-auto-pmp7233-bs-en
fasmicro, that's correct, the Cadence tool is very powerful, I utilized the Montecarlo corner simulation tool and I realized the strenght of this software.
My point is that if your budget cannot afford a expensive tool , you have to maximize the ratio between the power of the simulator utilized and the cost of the software , supposed the effectiveness of the design. That's not an easy equation to solve.
Agreed. A big company for sure can afford the expense of a tool like Cadence, a good free software however gives a chance to a small company to be a successful startup. This is not obvious but it requires the presence of a good engineer which has to be able to use all the potential strenghts of the free software, to realize an effective design.
I think that LTSPICE is really a useful software for a student or a young designer to increase his knowledge of electronics, and can be utilized for basic blocks of a complex integrated circuit. I think that it's all a matter of budget.
Then, PCB board will be tested under real data interacting with real environment
There is a report few years ago that showed that investing in simulators save more than 60% of product development costs. Provided the simulator is good, it is the best path to creating new products.
coolspice, I have not heard of that one. Spice simulators seem to be everywhere these days. The best about Spice is not the BSIM parameters but the environment the simulation is done. Cadence offers the best possible path to world-class sim IC design experience in my own opinion.
Among so many engineering tools, Cadence PCB design tool also provides auto route and generates a variety manufacture format with etch editing.
The leader is Altium Designer which used to be called Protel. I think Altium makes better PCB CAD than Cadence. Cadence is best in IC design.
Good summary @Chirshadblog,
1. Cost factor
2. Compatibility factor
3. Reliability factor
4. Cross funcional facor
We pay a lot of attention on the compatibility factor. From there we draw out Usability. It has to be usable for it to make sense. There are some CADs that are very difficult. If the CAD stays on the way to the actual job, reconsider its value.
A good engineer has to choose the tool taking into account the budget available and of the level of accuracy required for the product under design.
Simply look at your ROCE – Return on Capital Expenditure or ROI – return on investment. If the tool does not enhance your metrics, forget it. It is like the high school mechanical advantage where load must be bigger than effort, simulators must add value because it is an expense. That explains why you do not have to buy the best in the industry, consider your size and your need, a cheaper one may be good enough. After all, we do not all drive Bentley even though it is a better car to most cars. It is ROI.
>> And then, once the system is passed in the simulation environment, in real environment, the product will be reevaluated
That is the pattern and the reason why the CAD sector has grown over the years. Most products are entirely modelled in the SIM environment. The implementation validates the model. Yet, only a really good CAD can help make the real silicon to be close to the simulated one.
Candence's PSpice simulation technology is an advanced technology and industry-proven
Any idea what that costs. It turns out that Cadence will not even give you price unless they have a site visit where you will use the thing. I was trying to have price for some comparisons for analysis but not to make a purchase.
@goafrit2: you're correct, the ROI is the key parameter to take into consideration when purchasing a simulator, but sometimes an engineer has a further issue: to choose between two or more softwares having the same price. In this case the choice has to be done based on which simulator best fits the application requirements, it is not always an easy choice.
@etna powers: I would like to add further, that we need state-of-art tool for present set of applications. I have used LTSpice, Orcad-Pspice, they are good, but when it comes to analysis of the output of a simulation the free tools lag behind. We may not have to waste time in building a wrapper for a free tool.
The ROI is very important for us to sustain, but if we want to excel in the market, good tools are necessary. A good engineer can make good judgements and propose a confident design/product to the customer.
>> In this case the choice has to be done based on which simulator best fits the application requirements, it is not always an easy choice.
One major factor is to also consider the major software used in major universities. It is easier to align with these schools as you can get five years of free software training when you adopt the ones your future employees are already familiar with
>> The ROI is very important for us to sustain, but if we want to excel in the market, good tools are necessary.
That is not debatable. But when you are faced with dropping $10 million for Cadence license, you may have to look at ROI and ROCE to know if you can ask your startup funders that you need that type of expense in a new untested venture.
@goafrit2: I agree with you. That's the main reason for why a good interaction between colleges and companies is very important.
@amrutah, a good free tool will be developed only in case of a economic return, let's think for example to the linux software diffusion: it's free and powerful, many users like it. It is a good example of free software which works very well. Many programmers have developed further functionalities for this operating system.
The simulation tool that you cited are very good for a rough evaluation of the circuit, I have utilized in the recent past the LTSPICE for the evaluation of a test board and it was very useful to me.
I think that when the circuit to be simulated is very complex, for a better confidence of the designed circuit a not free software is required, keeping the balance between the ROI and the design confidence.
@etnapowers: Yes true but that is not easy as we think it is. There are internal conflicts which blocks many so this is one area of it.
@etnapowers >> I think that when the circuit to be simulated is very complex, for a better confidence of the designed circuit a not free software is required .
There is a number of areas where that holds true. Perhaps it is due to the fact that accurate simulation requires access to information which is often restricted from public domain.
The internal conflicts you're referring to are often generated by many different departments of a company, each department pushes according to its needs. For example, the purchasing people want a free software and the accuracy is not a primary requirement, on the other side the designer team want a accurate software, even if expensive, I think that the solution is a good balance between these two needs. For sure this analysis is not exaustive, but it gives a good idea of the blocks that you described, right?
>> That's the main reason for why a good interaction between colleges and companies is very important.
And the very foundation of great universities and the not-great ones. In some of the top colleges, research brings the industry into training and eductaion.
>> a good free tool will be developed only in case of a economic return, let's think for example to the linux software diffusion: it's free and powerful, many users like it
Still surprised why the creator of Linux made it free when Bill Gates was top of Forbes list. Now they use his idea to build IBM, Accenture and many EDA companies. Linux should have commercialized and made money out of his idea.
>> There is a number of areas where that holds true. Perhaps it is due to the fact that accurate simulation requires access to information which is often restricted from public domain.
When you have about 200k transistors, I have no clue how LTSPICE can help you. It will become a diminishing returns despite the cost factor. Compared to Cadence, there is no match. Cadence gets you going into anything the innovation can throw at you on transistor count or complexity.
>> For example, the purchasing people want a free software and the accuracy is not a primary requirement, on the other side the designer team want a accurate software, even if expensive,
That should be a really poorly managed company. In our firm, the engineers make the call on the CAD they will use. The purchasing dept only has to pay. Of course, we have to budget for it ahead.
>> any idea what that costs.
$1M for a location setup!
Agreed, that's the key for a new college to become a great university: to establish a deep relationship with the industry , a mutual relationship is recommended.
And the very reason why US and European universities are the best. They are structured to serve the industry and help drive innovations. The silos that exist within most developing world universities are not seen in these top schools that educate and research to change the world and industry.
@goafrit2: I couldn't agree more with you on this point. The US and European educational system is really effective in creating a deep link between the theoretical knowledge and its practical applications to the real world. I have been a student and I had the chance to apply the notions that I've learned from books into a industry environment where I had the chance to be the author of a Patent. It's really a great opportunity for a student, to grow and to became a professional.
> I have been a student and I had the chance to apply the notions that I've learned from books into a industry environment where I had the chance to be the author of a Patent
That educational system is the difference between these countries and the developing ones. While students in Africa, India etc are memorizing theories to pass exams by some lazy professors, the ones in U.S. are challenged to solve the next big problem in the society. The experience from such education can never be the same.
“The experience from such education can never be the same. “